DEV Community: Shilika

Will AI Replace Entry-Level PR Jobs? What Juniors Should Do Now

Shilika — Sun, 08 Nov 2026 05:13:37 +0000

AI won't remove entry-level PR jobs, but it is removing many of the tasks those jobs were built around: building media lists, clipping coverage, first-draft releases and monthly report formatting. The junior roles that survive will be smaller in number and different in shape, with less volume work and earlier exposure to judgment, writing and reporter contact. If you're starting out, the useful move is to practise the parts of the job a tool can't do, deliberately and early.

I'm writing this for juniors, and for the people who hire and train them. Neither group benefits from panic.

Which junior PR tasks is AI actually taking over?

Be specific, because "AI is taking jobs" is too vague to act on. Most junior PR roles were a bundle of tasks, and the bundle is coming apart.

Task juniors traditionally did	What AI does with it now	What's left for a person
Building media lists from scratch	Pulls reporters and recent articles by beat in minutes	Judging which reporter genuinely fits this story
Coverage clipping and monitoring	Tracks mentions and summarises them	Spotting the one mention that needs a response
First drafts of releases and pitches	Produces a clean draft from bullet points	Finding the angle, checking facts, making it sound human
Monthly reports	Formats metrics and coverage lists	Explaining what the numbers mean and what to change
Researching a reporter's past work	Summarises their last 20 articles	Reading enough to know what they'd actually care about
Repurposing content for social	Turns one piece into many formats	Choosing which version fits which channel

Look at the right-hand column. Every remaining task is a judgment call. That's the new entry-level job.

What does that mean for the number of junior roles?

Honestly, some agencies will hire fewer juniors, because one person with good tools can cover what used to take two or three. I won't pretend otherwise. If your plan was to spend two years doing nothing but media lists and clipping, that path is narrowing.

But two things push the other way. First, every company now produces more content, and someone has to check it, which needs people who understand media. Second, the senior people everyone wants to hire had to start somewhere. An industry that stops training juniors runs out of seniors. The firms that think long term will keep hiring juniors and give them better work earlier.

For more on why the senior end of comms is getting more valuable, not less, see my field note on whether AI will replace PR.

How did juniors learn the job before, and what's lost?

The old system had a hidden benefit. Building a media list by hand taught you who covers what. Clipping coverage every morning taught you which stories ran and why. Drafting twenty pitches that got rewritten taught you what a good one looks like.

That's the real risk with AI: not that it takes the tasks, but that it takes the learning that came with them. A junior who has never read a reporter's articles, because a tool summarised them, won't develop the instinct that makes senior people valuable.

I learned most of what I know by doing the volume work and paying attention. My advice when I wrote about it in 2023 still stands:

my advice to anyone starting out is - just learn by doing & then see yourself slay.

Shilika Jain on X, 7 Aug 2023

The doing has changed. The principle hasn't. You have to find a way to do the reps even when a tool could skip them for you.

The Weekly Reps Plan for junior PR people

This is the practice routine I'd give anyone in their first two years. It replaces the learning the old grunt work used to provide, on purpose.

Read five stories a week from one reporter on your beat, fully. Not summaries. Note what they lead with, who they quote and what they never write about.
Write one pitch a week without AI, then compare it to an AI draft. Find what yours has that the AI version lacks. Usually it's a specific detail or a reason why now.
Fact-check one AI draft line by line. Find the source for every claim. You'll quickly learn where models invent things, which is a skill teams now need.
Answer every journalist request you're given within the hour. Speed and reliability are how reporters start to trust a name they don't know.
Sit in on one interview or prep call. Ask to listen. Notice how the senior person steers the conversation.
Write one short opinion piece a month under your own name. A LinkedIn post, a newsletter, a guest blog. Writing in your own voice is becoming rarer and more valuable.
Ask one senior colleague "why did you do it that way" each week. The reasoning is what you can't get from a tool.

Point four comes from something I believe about the basics:

responding and answering journalists’ query is PR 101!

Shilika Jain on X, 25 Oct 2023

It sounds small. In practice, the junior who answers reporters fast and accurately becomes the person reporters email directly. That relationship is something no tool can take from you.

Which skills should juniors build first?

If I had to rank them for someone starting in 2026:

Writing without help. Clear, specific, short. AI makes average writing free, so good writing stands out more.
Fact-checking and source discipline. Teams that use AI need someone who catches what it gets wrong.
Reporter knowledge on one beat. Know the 20 to 30 people who cover your sector and what they care about.
Using AI tools well. Prompting for research and drafts, and knowing when not to trust the output.
Sector knowledge. In Web3 and AI, understanding the technology is the price of entry. You can't pitch what you can't explain.

That last one is how I started. The work was demanding in a way that taught me a lot:

love my job although it’s a mental stretch to learn about complex web3 tech,send 100s of emails,follow up w clients at 2AM in the night,break my head in devising new strategies,feeling blank due to too much information overload

Shilika Jain on X, 30 May 2023

AI can lighten the "100s of emails" part now. The "learn about complex tech" part is still yours to do. If you're specifically aiming at Web3, my sibling piece on how to break into Web3 communications covers the job search side.

What should managers and agencies do differently?

If you hire juniors, the old training model doesn't work anymore. You can't rely on volume work to teach them, because the tools do it. So build the learning in deliberately:

Old way juniors learned	New way to teach the same thing
Building lists by hand	Have them audit an AI-built list and explain each cut
Drafting endless pitches	Have them rewrite AI drafts and defend each change
Clipping coverage daily	Have them write a weekly "what this coverage means" note
Shadowing occasionally	Bring them into prep calls and debriefs as standard
Formatting reports	Have them present one insight from the report to the client

This also makes juniors useful faster. Someone who can check AI output, spot a weak claim and explain coverage to a client is worth more than someone who can only produce lists.

If you're a founder with a junior marketer and no senior PR person to learn from, pairing them with a fractional operator is one way to get both the work and the training. It's the kind of engagement I offer through AI startup PR.

Where is this advice wrong for you?

A few honest exceptions:

In very small teams or solo setups, there may be no junior role at all. Your path in might be freelancing, content writing or a role adjacent to PR, like community or social, and moving across.
In some regions and languages, AI tools are weaker, and native-language media work still relies heavily on people. Juniors with strong regional language skills have an edge there.
In highly regulated sectors, the review and compliance side grows, and juniors may spend more time on accuracy checks than on pitching. That's still valuable training.

And if you love writing but hate pitching, PR may not be the only home for you. Content and editorial roles are changing too, but they reward the same original writing skill.

FAQ

Should juniors avoid using AI tools so they learn properly?

No. Use them, because every employer will expect you to. Just don't let them replace the reps that build judgment. Draft with AI, but also write some things from scratch, and always read the source material yourself before trusting a summary.

Is PR still a good career to start in 2026?

For people who like writing, people and fast decisions, yes. The work is moving toward strategy and relationships earlier in a career, which is the more interesting part. It's a weaker choice if what you wanted was predictable, repetitive work.

What should I put in a junior PR portfolio now?

Show judgment, not volume. Include a pitch you wrote and why you targeted that reporter, an AI draft you corrected with your edits marked, and a short piece of original writing under your name. One placement you helped land, with your role explained honestly, beats ten you can't explain.

Will AI replace PR interns before it replaces junior staff?

Intern tasks are the most exposed, because they're often pure volume work. Good internship programmes will shift toward observation, writing practice and fact-checking. If an internship offers only list building, ask what else you'll learn before you accept it.

Training juniors on an AI-assisted team, or starting out yourself? Book 30 minutes with me.

I'm Shilika Jain, a fractional PR lead for Web3 and AI founders. More playbooks at shilikajain.com.

Twitter Threads vs Single Tweets: When Each One Wins in Crypto

Shilika — Sun, 08 Nov 2026 05:13:29 +0000

Use a single post when you have one idea that stands on its own: a reaction, an opinion, a sharp observation. Use a thread when the reader needs a sequence to follow, such as an explainer, a timeline, a step-by-step breakdown or a teardown, and when every post in it earns the next one. Most crypto content is better as a single post. Threads win for a narrower set of jobs, and they win big when they're doing those jobs.

I've written both for years. The two posts I still list in my portfolio are both threads from 2022: one on the battle of L1s and L2s, one on the FTX, Genesis and Gemini saga. Most of my day-to-day posting is single posts. Below is how I decide which one a given idea should be.

What's the actual difference between a thread and a single post?

A single post is one idea, delivered whole. The reader gets it, reacts or doesn't, and moves on. It's built for reach and reaction.

A thread is a reading experience. The first post has to win a click, and every post after it has to keep the reader going. It's built for depth and saves, and it asks more of the reader in exchange for giving more.

That trade is the whole decision. If your idea doesn't need the extra space, a thread just makes the reader do more work to get the same thing. If it does need the space, cramming it into one post either loses the nuance or turns into an unreadable block.

If you have access to longer posts on X, a single long post can sometimes replace a thread. The same test applies: does the reader need a sequence, or one point.

The thread test: five questions before you hit "add post"

I run any idea that's tempting me toward a thread through five questions. If I can't say yes to at least three, it becomes a single post.

Does it have a natural order? Steps, a timeline, a cause-and-effect chain, or a comparison point by point. If you could shuffle the posts without losing anything, it's a list, and lists usually work better as one post.
Would a reader save it to come back to? Explainers, frameworks and resource lists get saved. Opinions mostly don't.
Can every post stand as a useful sentence on its own? Each one should hold a fact or a step, not filler like "but wait" or "here's where it gets interesting".
Is there enough evidence to fill it? A thread padded to seven posts with two real points in it costs you trust.
Is the first post strong enough to stand alone? Most people only ever see the first post. If that post isn't worth reading by itself, the thread fails before it starts.

Which crypto content works as a thread, and which as a single post?

Here's the split I work from, by type of content.

Content type	Better as	Why
Reaction to breaking news	Single post	Speed matters and reporters lift single posts more easily
Hot take or opinion	Single post	One claim lands harder without padding
Protocol or concept explainer	Thread	Readers need a sequence to build understanding
Timeline of an event (exploit, collapse, governance fight)	Thread	Chronology is the structure
Marketing or launch teardown	Thread, or single post with image	Depends on how many distinct points you have
Resource list (tools, podcasts, newsletters)	Thread, or single post if short	Each item can carry a line of context
Lesson from your work	Single post	One lesson, told well, travels further
Comparison of two projects	Either	A thread for many dimensions, a single post with a visual for a few

The explainer and timeline rows are where threads earn their keep. When I wrote about L1s and L2s in September 2022, the topic genuinely needed steps.

1/ 🤺The smart contract wars are on fire, here's a deep understanding of L1's v/s L2's:

A Thread [🧵]

Shilika Jain on X, 5 Sep 2022

The next day I followed it with a second thread on why L2s exist at all, built around the blockchain trilemma.

Why have L2's when you have so many L1's:

Understanding how The Blockchain Trilemma:

A THREAD (🧵)

Shilika Jain on X, 6 Sep 2022

Two threads on consecutive days worked because each answered a different question. Merging them into one long thread would have lost readers halfway.

How should you structure a single post?

A good single post in crypto has three parts, and it can do all three in under 280 characters:

The claim. What you think, said plainly in the first line.
The proof. One specific detail: a number you can stand behind, an example, a pattern you've seen.
The edge. What it means, or who should care.

A worked example, using a lesson from my own client work. A weak single post: "Launches need a strong narrative! Think carefully about your messaging." A strong one: "Your mainnet launch doesn't need ten features in the pitch. It needs one angle. Rari Chain led with NFT creator royalties protected at the chain level and landed in CoinDesk, The Block and Cointelegraph the same week."

The strong version has a claim, proof the reader can check, and an instruction. It doesn't need a second post.

Single posts also work for comparisons when you can show rather than tell. In 2022, when a couple of threads were already comparing Aptos and Sui on the tech, I posted the marketing side as one post with an image instead.

I've seen a couple of threads differentiating between @AptosLabs & @Mysten_Labs on a technical standpoints:)

But here's the specs from the marketing side👇

Shilika Jain on X, 21 Sep 2022

The image carried the comparison. A thread would have split something that worked better seen at once.

How should you structure a crypto thread?

A thread has more moving parts, so it needs more structure. The shape I use:

Post 1, the hook. States the topic and the payoff, and stands alone. Number it "1/" or mark it with 🧵 so readers know there's more.
Post 2, the context. Why this matters now. One post, not three.
Posts 3 to 7 or so, the body. One idea per post, in order. Each should make sense if someone lands on it from a repost.
A visual somewhere in the body. A diagram, a table screenshot or a chart breaks the wall of text and gets saved.
The closing post. A one-line takeaway and, if it fits, a pointer to something longer you've written.

Keep most threads between five and ten posts. Past that, I'd ask whether it should be an article instead, and that's often the right call: a strong thread is frequently the first draft of a blog post. I go through that process in turning X posts into blog and LinkedIn content.

Things that kill a thread: a hook that's all suspense and no substance, posts that exist only to build tension, and an ending that asks for a follow without having earned one.

Do threads or single posts get more reach on X?

It depends on the post, not the format, and anyone who gives you a fixed answer is guessing about an algorithm that keeps changing. What I can say from practice is that each format earns a different kind of attention.

Single posts are better at quick reactions, replies and reposts, because they're easy to engage with in a scroll. Threads are better at saves, profile visits and follows from people who want more of your thinking, because they show depth. A balanced account needs both. Mostly single posts to stay present, with a thread when you have something that needs the room.

For founders, I'd frame the mix around what each format does for your profile. Single posts keep you in the conversation and make you quotable to reporters, which I cover in how journalists find sources on X. Threads build the record that convinces an investor or a hire you understand your market. My 12-month X operating system for Web3 founders shows how the two fit into a weekly plan.

When does a thread hurt you?

When it's a single idea wearing a costume. Readers can tell when a thread exists for format reasons, and it costs you trust that a plain post wouldn't.

Threads also hurt during fast-moving news. If an exploit is unfolding, a seven-post thread written in a rush will contain mistakes, and the first post will be screenshotted with them. One careful post, followed later by a thread once the facts are clear, is safer.

And they take time. A good thread takes me far longer to write than a single post, because each post has to be edited to stand alone. If you're posting while running a company, spend that time on one thread a week at most, and let single posts carry the rest. If you'd rather have help shaping long-form threads and the articles they turn into, that's part of my content writing work.

FAQ

How long should a crypto thread be?

Long enough to cover the sequence and no longer. Most of mine land between five and ten posts. If you find yourself going past twelve, the topic probably wants to be an article, with a short thread that points to it.

Should I number the posts in my thread?

Numbering ("1/", "2/") helps readers know where they are and signals a thread in the first post. Some people prefer only a 🧵 on the first post to keep the rest clean. Either works, as long as the first post makes clear more is coming.

Can I turn an old single post into a thread later?

Yes, and it's a good way to develop ideas. If a single post got strong replies or questions, that's a sign people want more. Quote your original post and expand it into a thread that answers what readers asked.

Is it better to post a thread all at once or one post at a time?

Post it all at once. Drip-feeding a thread over hours means early readers see an unfinished piece and many won't come back. Write it in a draft or a scheduling tool first, check that each post stands alone, then publish the whole thing.

Have an idea that's either a great thread or a great article and can't decide which? Book a 30-min teardown with me and we'll shape it together.

I'm Shilika Jain, a fractional PR lead for Web3 and AI founders. More playbooks at shilikajain.com.

How Indian Crypto Startups Get Covered in US Media

Shilika — Sat, 07 Nov 2026 05:14:08 +0000

Indian startups get covered in US media when the story stops being "an Indian startup did something" and becomes a category story a US reader already follows, backed by proof that holds up outside India. In practice that means winning Indian business press first, then reframing for global crypto and tech desks with a category angle, a comparable metric and a founder who can comment on the US news cycle. Origin is rarely the news. What you're adding to the category is.

I'm an Indian PR consultant who also writes, as a freelance author for The Chainsaw and a contributor at CryptoDailyUK. I described myself on X this way a few weeks ago:

pr consultant and journalist from India ✨

Shilika Jain on X, 17 Sep 2026

So I see this problem from both sides of the inbox. Indian founders send global desks the pitch that worked at home, and the global desk can't see a reason to run it.

Why US media passes on most Indian startup news

Indian business press and US crypto press are answering different questions. An Indian outlet asks what this means for Indian founders, Indian capital and Indian users. A US crypto desk asks what this means for the category: payments, DePIN, AI agents, stablecoins, gaming. "Indian" is a strong hook at the first desk and almost no hook at the second.

Three things usually go wrong in the pitch itself:

The pitch leads with geography. "India's first" tells a US reporter the story is local.
The numbers don't travel. User counts, revenue in rupees and India-only partners are hard for a reporter to compare with the companies they already cover.
The founder is unavailable in US hours. By the time the reporter replies, it's past midnight in Bengaluru, and the window closes.

I wrote this about launches in general, but it lands hardest on cross-border pitches:

every other launch is important to the company & promotional to the media of it doesn't have a high impact value associated w it!

Shilika Jain on X, 11 Oct 2024

High impact value, for a US desk, means impact on something its readers already track.

What do Indian and US outlets each reward?

This is how I brief Indian founders before we write a single global pitch. It's a pattern from the campaigns I've run, not a rule every outlet follows.

What the reporter weighs	Indian business and startup press	US and global crypto or tech press
The lead	Indian founder, Indian round, Indian market	A category shift, a first, a conflict or a trend
Proof	Round size, investors, Indian users	Usage, revenue, on-chain data, named partners readers recognise
The founder	Origin story and ambition	A clear thesis and a view on the week's news
Timing	Indian working day	US morning, tied to what's already in the news
Typical outlets	Economic Times, Inc42, YourStory, Entrackr	CoinDesk, Cointelegraph, Decrypt, The Block, Blockworks
Format that wins	Funding story, founder profile	Exclusive, data story, op-ed, expert comment

If you only read one row, read "The lead". Most Indian pitches that fail abroad are good Indian pitches sent to the wrong column.

The four swaps that turn an India story into a global one

When an Indian founder wants global coverage, I don't start by rewriting the release. I make four swaps in the story, and then I write.

1. Swap geography for category. Take a made-up example: "Bengaluru startup raises $5M" becomes "a DePIN network bets that idle office GPUs can undercut cloud prices for AI inference." The city moves to the second paragraph.

2. Swap local numbers for comparable ones. Replace rupee figures and India-only metrics with numbers a US reporter can compare: transactions, active wallets, retention, nodes, developers. If you have on-chain data, use it. My notes on turning on-chain data into story hooks go deeper on this.

3. Swap the origin story for a thesis. The Indian profile is about how you started. The US interview is about what you believe the category gets wrong. Pick one sentence you'd defend on a podcast and repeat it everywhere.

4. Swap the announcement for commentary. Your first global placement often isn't your own news. It's a quote on someone else's. Offering a sharp view when a big story breaks gets an unknown founder into outlets that wouldn't run their funding news. I cover how to run this in reactive commentary for Web3 and AI founders.

A quick test before you pitch: delete every mention of India from your first paragraph. If the paragraph still has a reason to exist, it's ready for a US desk.

What the Bullieverse raise shows about the two tracks

Bullieverse, an Indian-founded metaverse gaming team, raised $4M in February 2022. The round ran on two tracks with the same facts.

In India it was a mainstream business story. Economic Times ran it as Bullieverse bagging $4 million in a round led by Web3 venture funds, and it also landed in Inc42, Entrackr, YourStory, VCCircle and Tech in Asia. No reframing needed. An Indian team raising from Web3 funds was the story.

The global side was different. It ran as a press release on Cointelegraph, on NewsBTC, as a founder interview on Hackernoon and on Investing.com, where the hook was the Chainlink VRF integration rather than the round. The CoinDesk piece was sponsored content, not earned, and I'd rather say that plainly than let it read as a tier-1 win.

The lesson I took from it: the Indian press came for the founders and the money, and the global tech pickup came for the integration, which was a category detail. The Bullieverse case study has the full list.

Build local, promote global: the sequence I use

I've used this line for years and it's still how I'd order the work:

Web3 funda: BUIDL local, promote global.

Shilika Jain on X, 6 Dec 2022

For an Indian crypto founder, "promote global" works best as a sequence over a quarter, not a single launch day:

Win the Indian business press first. Economic Times, Inc42 and YourStory give you credibility you can point to, and they're the right audience for hiring and Indian partners. The India Web3 PR playbook covers this step.
Collect proof that travels. Usage numbers, on-chain data, a named partner a US reader recognises.
Build the founder's English-language footprint. Regular posts on X and LinkedIn around one thesis, so a reporter who looks you up finds a point of view.
Start with commentary. Two or three expert quotes in global crypto outlets before you ask anyone to cover your own news.
Pitch the next real milestone globally, with the category angle on top and an exclusive offered to one outlet.
Be reachable in US hours. Block two evenings a week, Indian time, for calls in the launch fortnight.

Step 6 sounds small. It's the one founders skip most, and it costs them the interview that was one reply away.

When chasing US media is the wrong goal

US coverage isn't always worth the effort, and the cost is real: a reframed narrative, weeks of commentary work and late nights.

Skip it, or push it later, when:

Your users are in India. An Economic Times story reaches the people who'll download your app. A CoinDesk story mostly reaches people who won't.
You don't have comparable proof yet. A global pitch without numbers a US reporter can check usually ends as a sponsored post, which won't do much for your credibility.
Your founder can't commit to the time. Commentary only works if you're quick, and quick means available in US hours.

Go after it when you're raising from global funds, preparing an exchange relationship, hiring internationally or when your category is being decided in US media right now.

If you want help planning the order, my founder profiling service is built around exactly this kind of sequence.

FAQ

Do Indian founders need a US-based PR agency to get US coverage?

No. What you need is someone who knows which global crypto and tech reporters cover your category and how they like to be pitched. Location matters less than relationships, the angle and being reachable in US hours.

Is paid or sponsored coverage in US crypto outlets worth it for Indian startups?

It can help with search visibility and a listing on a well-known site, but it doesn't carry the credibility of earned coverage, and reporters can tell the difference. Treat it as distribution, label it honestly in your own reporting and don't build your story around it.

How long does it take an Indian crypto founder to get a first US placement?

Expert commentary can land within a few weeks if the founder has a clear view and answers quickly. A feature or exclusive on your own news usually takes longer, because it needs proof and a category angle the reporter can sell to their editor.

Should the founder mention being from India in global pitches?

Yes, where it adds something, like access to India's developer base or Indian users as proof of demand. Just don't make it the headline unless the story is specifically about India.

Indian founder aiming at CoinDesk or Decrypt next quarter? Book a 30-minute call and I'll show you which of the four swaps your story still needs.

I'm Shilika Jain, a fractional PR lead for Web3 and AI founders. More playbooks at shilikajain.com.

Incentivized Testnet Marketing: A Comms Plan for Real Testers

Shilika — Sat, 07 Nov 2026 05:14:01 +0000

Incentivized testnet marketing works when you describe the job before the reward: what you need tested, what good testing looks like, and what you will never promise. If the first thing people learn is that participation "may be rewarded", you'll attract farmers and spend the next three months managing their expectations. Publish the purpose, the rules and the limits up front, and the testers worth keeping will self-select.

Why do incentivized testnets turn into farming campaigns?

Because most teams market them as a pre-airdrop. The announcement thread leads with points, the community manager hints at "early contributor recognition", and a dozen airdrop-hunter accounts write their own guides within a day. From that moment, the people arriving are optimising for eligibility, not feedback.

The incentive is not the problem. The order of information is.

When the reward is the headline, every later message gets read as a clue about the reward. A bug-bounty update becomes "do bugs count for points". A node requirement becomes "how many wallets can I run". Your actual goal, finding the edge cases before mainnet, slides to the bottom of the conversation.

A testnet is the cheapest place to find out which crowd you're building for. It's a question I keep coming back to.

but q's that keep running on my mind:

are projects going to be sustainable w/o enuf revenue?

are users in for incentives v/s use cases

Shilika Jain on X, 22 Aug 2024

Your testnet comms answer that second question before mainnet does. Airdrop hunters will show up whether you like it or not, so plan for them as an audience, and a plan that pretends they don't exist will be written for an imaginary crowd.

What should you promise in testnet comms, and what should you never promise?

This is the part teams skip, and it's the part that creates backlash later. Before anyone writes a tweet, I want a one-page promise ledger: a short table that sorts every claim you might make into three columns, signed off by the founders and whoever owns legal.

Topic	Safe to say	Never say	Why it matters later
Rewards	"Contributions are being recorded" (only if true)	"Testers will get the airdrop"	A soft hint gets screenshotted and treated as a commitment
Criteria	The behaviours you value: bug reports, uptime, feedback quality	Exact thresholds or formulas	Published thresholds get scripted within days
Timeline	Testnet phases and their end dates	Token launch dates	You can control phase dates; you can't control market timing
Sybil policy	That multi-wallet activity will be filtered, and that you'll explain the method after	That nobody will be excluded	Exclusions are the most common source of post-drop anger
Changes	That rules can change, and how you'll announce changes	Silent edits to the docs	Undocumented changes read as moving the goalposts

The "never say" column is the valuable one. Every line in it is something I've seen communities quote back at teams after a distribution. If your plan needs one of those sentences to get sign-ups, the plan is buying wallets, not testers.

The five-phase incentivized testnet comms plan

Here's the sequence I'd run for a testnet with a reward component. The timings are logic, not data, so stretch them to fit your build schedule.

Brief the job (about three weeks before opening). Publish a short post on what you're testing and why it matters for mainnet. Name the tester profiles you need: node operators, developers integrating your SDK, users trying the core flow on a phone. No reward language yet.
Open with the rules (launch day). Release the testing guide, the feedback channels, the sybil policy and the promise ledger's safe lines in one place. One canonical page beats ten threads.
Report what testers found (weekly). A changelog that credits specific bug reports by handle does more for tester quality than any leaderboard. It shows the crowd what "valuable" looks like.
Announce changes before they happen (as needed). If you adjust criteria or extend a phase, say so at least a few days ahead and explain the reason. Never edit the docs quietly.
Close with a public wrap-up (within two weeks of ending). Share what the testnet changed in the product, how many issues were fixed, and how contributions will be evaluated. If there's a reward, this is where the method goes, not the allocation. When the eligibility announcement itself comes, I cover the wording in announcing airdrop eligibility without a farming narrative.

Phase three is the one founders underrate. A weekly "what testers found" post is earned content, it gives your engineers a reason to talk publicly, and it quietly tells farmers that clicking buttons isn't what you're counting.

How do you write the announcement so it attracts testers?

Lead with a problem only a real tester can help with. "We need 200 people running the light client on low-bandwidth connections in Southeast Asia" filters better than "Join our incentivized testnet". It's specific, it implies effort, and it invites people who actually have that setup.

A few rules for the announcement itself:

Put the testing goal in the first line and the reward note, if any, near the end.
Show a real example of a useful bug report, so people can see the standard.
Say plainly that multi-wallet activity will be filtered.
Link one guide, not a quest board with twenty tasks.
Give a safety note about wallets and fake links.

That last point deserves its own line. Testnets attract phishing sites, fake faucets and impersonator accounts, and the people chasing rewards are the most likely to click.

people running behind airdrops, just make sure that you don't end up getting your wallets compromised!

Shilika Jain on X, 23 Mar 2023

Pin a short security notice: the only official links, the fact that you'll never DM first, and that testnet tokens have no value. It protects users and it signals a team that thinks about them as people.

Should you pitch an incentivized testnet to the press?

Usually not on its own. "Project opens incentivized testnet" is a crowded headline in crypto media, and reporters know the participant counts are inflated by farming. The wrap-up is a better story than the launch.

What journalists can use is what the testnet proved. A shift in throughput under real load, a security issue caught before mainnet, an unexpected region where node operators clustered. Those are findings, and findings can be pitched. If you have on-chain data behind them, my post on on-chain data as PR story hooks covers how to turn numbers into an angle.

Building in public is the posture that makes this work.

every web3 wants to onboard the next million/billion users in web3!

nobody actually elaborates on how they'd do it!

Shilika Jain on X, 10 Jul 2023

A testnet is one of the few moments you can show the "how" with evidence. Teams that report honestly on what broke tend to get a warmer reception at mainnet than teams that only post sign-up milestones. I fold this into launch sequencing as part of my token launch PR work, because the testnet wrap-up often becomes the first credible proof point in the TGE narrative.

How do you measure whether the testnet attracted real testers?

Ignore registered wallets. Look at the work.

Signal	Mostly farmers	Mostly testers
Bug reports	Few, duplicated, copied from guides	Specific, reproducible, with device or environment details
Activity curve	Spikes after each reward hint	Steady, with bumps after product releases
Support questions	Eligibility, snapshots, "is this counted"	Errors, configuration, documentation gaps
Feedback channel	Empty except during announcements	Ongoing threads, testers answering each other
After the phase ends	Activity falls to near zero	A core group asks about mainnet and next steps

If the right column is thin, don't respond by adding more rewards. Shrink the scope, recruit directly from developer communities and existing users, and run a smaller, closed phase with people you've spoken to.

When doesn't this approach fit?

There are trade-offs. A testnet that downplays rewards will grow slower, and your sign-up chart will look modest next to a competitor's points campaign. If your goal is purely awareness before a token, you may decide farming traffic is an acceptable cost. That's a legitimate choice, as long as you don't later present those wallets as users. My sibling post on points program communications covers that model.

It also doesn't fit when the network genuinely needs volume, for example stress-testing throughput. Then you want lots of activity and can say so honestly: "we need load, here's the window, here's what we're measuring". Being clear about why you want the traffic is what keeps it from turning into a grievance.

And if you have no reward at all, this plan still works. You just cut the ledger's reward row and keep everything else.

FAQ

How long should an incentivized testnet run?

Long enough to cover at least one meaningful product release inside the window, so you see behaviour after a change and not only at launch. For many teams that means several weeks per phase. Shorter windows mostly reward speed, which favours scripted wallets.

Should we publish a leaderboard?

Only if it ranks something that's hard to fake, like accepted bug reports or verified uptime. Leaderboards based on transaction counts invite automation and give farmers a target. If you publish one, explain what it measures and that sybil filtering happens before any reward.

Can we change the reward criteria mid-testnet?

You can, but announce it ahead, explain the reason, and apply it going forward rather than retroactively where possible. Retroactive changes are what communities remember. Keep a dated changelog of every rule edit on the canonical page.

Do we need KOLs to promote a testnet?

Not for recruitment. Creators bring reach, and reach is the one thing a testnet doesn't lack. If you work with any, brief them on the testing goal and ask them to share the guide, not reward speculation.

Opening a testnet in the next quarter and want testers who stick around for mainnet? Book a 30-minute comms review before the announcement goes out.

I'm Shilika Jain, a fractional PR lead for Web3 and AI founders. More playbooks at shilikajain.com.

Does PR Generate Leads for B2B? Where Coverage Shows Up in Sales

Shilika — Fri, 06 Nov 2026 05:14:07 +0000

PR does generate some direct B2B leads, but far fewer than founders expect, and judging it on inbound form fills will make it look like a failure. Its real effect on a B2B pipeline is indirect: coverage builds trust before the first call, shortens the diligence a buyer does on you, answers objections your sales team keeps hearing, and gives your internal champion something credible to forward. If you measure it at those points, you'll see it working.

I've run PR for infrastructure, wallet, data and AI companies whose buyers are developers, exchanges, enterprises and other protocols. None of those buyers read one article and sign. They read it, remember the name, and then check you out three months later when a need appears. That's the shape of the effect.

Why doesn't coverage turn straight into leads?

B2B buying is slow, committee-driven and full of risk checks. A head of engineering who reads a Blockworks piece about your MPC wallet stack doesn't fill in your contact form that afternoon. They don't have budget yet, or they're mid-contract with someone else, or they need three colleagues to agree.

What changes is their starting point. Next time your SDR emails them, or they see your name in an ecosystem call, you're not a stranger.

before you even look to acquire users, work on positioning + increasing awareness!

Shilika Jain on X, 7 Dec 2022

That's the order of operations. PR sits upstream of acquisition, and a cold outbound sequence lands differently when the buyer has already seen you in a publication they respect.

Direct leads do happen. They tend to come from founder podcasts and long interviews, where a buyer hears the person they'd be working with think out loud for forty minutes, more often than from news stories. Count them when they come. Just don't build the business case on them.

Where does PR show up in a B2B deal?

Here's the map I use with founders to plan where coverage gets deployed. I call it the deal-stage coverage map: for each stage of a typical B2B cycle, what the buyer is worried about, what coverage does about it, and the asset your team should actually send.

Deal stage	What the buyer is thinking	What coverage does	Asset to use
First touch	"Who are these people?"	Turns a cold name into a recognised one	Best single piece of tier-1 or trade coverage, linked in the outbound signature
Discovery call	"Do they understand my problem?"	Shows the founder's point of view on the category	Founder op-ed or podcast episode on the problem, not the product
Technical evaluation	"Is this real or vapour?"	Independent confirmation that something shipped	Coverage of the launch, integration or partnership
Security or vendor review	"Will they still exist in two years?"	Signals funding, backers and staying power	Funding announcement coverage
Internal champion selling	"How do I convince my boss?"	Gives the champion third-party proof to forward	A short pack of three links plus a one-line summary each
Renewal or expansion	"Are they still the right bet?"	Reminds the buyer the company is growing and visible	Recent coverage and new founder content

Notice that almost none of this is about the reader of the original article. The article's main job in a B2B context is to exist, be findable and be forwardable.

How does coverage shorten diligence?

Every serious B2B buyer googles you. Increasingly, they also ask an AI assistant. What comes back is the first round of diligence, and it happens before you're in the room.

If the results are your own website, a press release on a wire service and a few social posts, the buyer has learned nothing independent. If the results include a CoinDesk Protocol newsletter mention, a Blockworks explainer and a founder interview, a lot of the "are they legit" questions are already answered.

Web3Auth is a good example from my client work. In one quarter its Google Cloud partnership, a Firebase extension, was covered by Benzinga, The Cryptonomist, Cryptonews FR and the CoinDesk Protocol newsletter, and Blockworks covered its MPC wallet security work. For a buyer evaluating wallet infrastructure, that body of coverage answers the "is this a serious team" question before a sales engineer has to. I can't tell you how many deals that touched, and I wouldn't invent a number. I can tell you that's the mechanism.

your tech isn’t your asset, it’s the people believing in it🥹

Shilika Jain on X, 31 Dec 2024

In a B2B sale, "the people believing in it" includes reporters, analysts and other customers whose belief is visible in public. That's what coverage makes visible.

Can PR handle sales objections?

Yes, and this is the most under-used application I see. Every sales team has three or four objections they hear every week. Some of them are really perception problems, and perception problems are PR problems.

The process I run:

Collect the objections. Ask your sales lead for the five objections that come up most, in the buyer's own words.
Sort them. Separate product gaps (no SOC 2 yet, missing a feature) from perception gaps ("you're too small", "you're crypto-only", "nobody else uses you").
Leave product gaps alone. PR can't fix a missing feature and shouldn't pretend to.
Assign each perception gap a story. "Too small" needs funding and customer stories. "Crypto-only" needs a mainstream or enterprise angle, like Fluence's DePIN story tied to a global Microsoft outage in e27. "Nobody uses you" needs a usage or data story.
Pitch those stories over the next quarter. Then hand the resulting coverage back to sales with a note on which objection each piece answers.

This turns PR into something a sales team actively asks for, instead of a vanity line in the board deck. If you sell into enterprise AI buyers, the deeper version of this alignment is in aligning press, analyst relations and sales for enterprise AI.

What does a good PR and sales handoff look like?

Most companies get coverage and then do nothing with it beyond a post on X. The handoff is where B2B value is captured or lost. A good one includes:

A coverage library your sales team can actually search, tagged by objection and buyer type, not a folder of PDFs.
One-line summaries for every piece, written so a rep can paste them into an email without editing.
A fortnightly note from PR to sales: what ran, what's coming, which piece to use for which deal.
Founder content in the outbound sequence. A podcast clip or op-ed often outperforms a product one-pager as a second touch.
Proof on the website. A press page and "as covered in" strip near the pricing or demo page, where buyers look when they're deciding.
A feedback loop. Reps tell PR which pieces buyers mentioned on calls, which shapes the next quarter's pitching.

Customer stories deserve their own mention. A journalist rarely runs a vendor's customer story as news, but a well-built case study that references independent coverage is one of the strongest sales assets you can have. The B2B case study template is a good starting structure.

How should you measure PR's effect on pipeline?

Not with last-click attribution. B2B journeys are too long and too offline for that. Use signals that sit closer to the mechanism:

Self-reported source on demo forms, with a free-text "how did you hear about us" field. Buyers will often write "saw you in Blockworks" or "heard the founder on a podcast".
Mentions on calls. Ask reps to log when a prospect references coverage unprompted.
Sales cycle length for deals where coverage was used versus not. This is directional, not proof, but it's worth watching.
Champion asks. Count how often a buyer asks for "something I can send my team".

There's a sibling piece on how to attribute signups to PR coverage that goes deeper on the tracking side. For the full measurement picture, how to measure PR results for startups covers the rest.

Your honesty might make you lose a client but will bring you a credible reputation forever 🤝

Shilika Jain on X, 4 Apr 2022

The honest version of this conversation, which I have with every B2B founder before we start, is that I can't promise a number of leads. What I can promise is that their sales team will stop starting every call from zero.

When is PR the wrong tool for B2B pipeline?

A few cases where I'd tell you to spend elsewhere first:

You have no sales motion yet. If nobody is following up on the attention, coverage evaporates. Build the basic outbound and demo process first.
Your buyer universe is tiny. If you sell to twenty exchanges, direct relationships and events will beat media coverage.
You need pipeline this quarter. PR's effect is cumulative. If the board needs meetings in the next 60 days, outbound and partnerships are faster.
Your product can't survive the scrutiny. Coverage sends serious evaluators your way. If the product isn't ready for them, you've just accelerated the bad first impression.

When those conditions aren't true, PR becomes one of the cheapest ways to make every other sales activity work a bit better. My Web3 PR campaigns and AI startup PR work is built around exactly this kind of sales-aware pitching.

FAQ

Should my PR agency be paid per lead?

No. Pay-per-lead models push agencies toward paid placements and promotional content that journalists ignore, and they reward the wrong outcome. Agree on placements in target outlets, message pull-through and a sales handoff process instead, then review pipeline signals together each quarter.

Is a press release on a wire service useful for B2B sales?

Only as a reference document. Wire releases rarely get read by buyers, and they don't carry the independent credibility of a reporter's story. If you need a link for the website, a release is fine, but don't count it as coverage or expect it to move a deal.

How many articles do I need before sales sees a difference?

There's no magic number, but a small, consistent set beats one big hit. Three or four pieces in outlets your buyers actually read, plus a couple of founder interviews, is usually enough for a sales team to start using them. What matters more is that they cover different objections.

Do podcasts work better than articles for B2B?

For trust, often yes, because a buyer gets to hear how the founder thinks for a long stretch. Articles are better for quick, forwardable proof. Most B2B programmes I run use both, with podcasts feeding discovery and articles feeding diligence.

If your sales team keeps starting every call by explaining who you are, let's talk about what PR should be doing for your pipeline.

I'm Shilika Jain, a fractional PR lead for Web3 and AI founders. More playbooks at shilikajain.com.

Crypto Explainer Video Script: A 75-Second Structure

Shilika — Fri, 06 Nov 2026 05:13:56 +0000

A good crypto explainer video script runs 60 to 90 seconds and follows five beats: a problem shot the viewer recognises, the turn where your product appears, one action shown on screen, one piece of proof, and a single next step. Write it at roughly 150 to 200 words for a 75-second cut. Leave out the architecture, the tokenomics and the roadmap. The viewer should finish knowing what changes for them, not how the chain works.

Most crypto explainers fail in the first eight seconds because they open with the company. Nobody has a reason to care about your company yet. They have a reason to care about their own problem.

Why do most crypto explainer videos lose viewers early?

They're written by the people who built the product, in the order the product was built. Consensus first, then the token, then the app, then, if there's time, a user.

That order makes sense to an engineer and no sense to a viewer. Someone watching an explainer on your homepage, in a conference booth loop or in a pinned post on X is asking one question: is this for me? If the first shot is a spinning logo and a line about decentralised infrastructure, the answer they reach is no.

Marketers in this space spend a lot of their week repeating the same explanation:

being a crypto marketer is hard:( you gotta attend conferences,distribute merch, have ppl complete quests, explain your product 101 times, think of creative campaigns, network/converse with industry players

Shilika Jain on X, 6 Dec 2024

A good explainer is how you stop doing the 101st explanation by hand. That only works if the video explains it better than you do in person, which means it has to be at least as plain as you are at a booth.

What is the best structure for a crypto explainer script?

I use a five-beat structure timed for 75 seconds. It stretches to 90 or compresses to 60 by adjusting the middle beats, never the opening.

Beat	Time	Job	Rough word budget
1. Problem shot	0 to 8s	Show a moment the viewer has lived	15 to 20 words
2. The turn	8 to 20s	Name the product as the fix, in one sentence	25 to 30 words
3. Show one action	20 to 45s	The product doing one thing, on screen	50 to 60 words
4. Proof	45 to 62s	One number, partner or result you can stand behind	30 to 40 words
5. Next step	62 to 75s	A single thing to do now	15 to 25 words

The word budget assumes a natural voiceover pace of about two to two and a half words per second, with room for a few silent shots. Don't trust my numbers over a stopwatch: read your draft aloud at the pace you'd actually record it and cut until it fits.

How do you write the opening problem shot?

Pick one person in one moment of friction, and show it before you say anything about yourself.

Good problem shots are specific and a little uncomfortable. A freelancer watching a transfer fee eat part of a payment. Someone staring at a 12-word recovery phrase with no idea where to keep it. A developer waiting on an RPC that keeps timing out. Bad problem shots are abstract: "The financial system is broken." "Web3 is fragmented." Those are conclusions, and viewers don't sit through conclusions they didn't reach themselves.

A quick test for your opening line: could someone who's never bought crypto picture it? If yes, keep it. If it needs a glossary, rewrite it.

The opening is also the only beat you should never shorten. When the cut runs long, take time out of the proof or the action, not the problem.

What should you show in the middle of the video?

One action. Not a feature tour.

Pick the single thing a new user does that delivers the outcome you promised in beat 2, and show it end to end on screen. Tap, confirm, done. If your product needs four screens to deliver its first useful moment, either show only the last two or fix the onboarding before you film anything.

The strongest explainer idea I've pitched recently wasn't a video at all. In my June 2026 teardown of the prediction market Limitless, I suggested a giant live "Future Board" at conferences, where attendees place predictions and watch the odds move. The reason it works is the same reason a good middle beat works:

Attendees place live predictions and watch probabilities move in real time on a massive screen. Every journalist walking by instantly gets the product.

Shilika Jain on X, 9 Jun 2026

Watching the product do its thing beats any description of it. Your middle beat should feel like that board: one visible cause, one visible effect. I unpack the rest of that idea in the Limitless marketing teardown.

If you're tempted to use an analogy instead of a demo in this beat, test it first. Some analogies collapse the moment someone asks a follow-up question, and I cover how to check that in analogies to explain blockchain products.

A worked example: a fictional 75-second script

Here's the structure filled in for a made-up stablecoin payments app called "Lumen Pay". It isn't a client and the details are invented to show the shape.

Beat 1 (0 to 8s). Close-up of a phone. A payment notification shows less than the invoice amount. Voiceover: "You did the work. Then the transfer fees took their cut, and it still took four days."

Beat 2 (8 to 20s). Voiceover: "Lumen Pay lets clients pay you in digital dollars. It lands in minutes and the fee is shown before anyone hits send."

Beat 3 (20 to 45s). Screen recording: client opens a payment link, sees the amount and fee, confirms. Cut to the freelancer's phone receiving it. Voiceover keeps pace with the taps: "Send a link. Your client sees exactly what they'll pay. One tap to confirm. You're paid. Cash out to your bank whenever you want."

Beat 4 (45 to 62s). One proof point on screen. For a real product this is a number you can defend, a named partner, or a regulatory status. Never a projection.

Beat 5 (62 to 75s). "Create your first payment link in under a minute. Link below."

Notice what's missing. No mention of which chain it runs on. No token. No team photos. The word "blockchain" doesn't appear once.

What should you leave out of a crypto explainer video?

This is my cut list. If any of these are in your script, they need a very good reason to survive.

Consensus mechanism and architecture. Belongs in docs and developer content.
Tokenomics. Belongs in the token page and investor materials. In a product explainer it reads like a sales pitch for the token.
A wall of partner or chain logos. Viewers can't read them at video speed. Pick one partner as proof in beat 4 or skip it.
The roadmap. An explainer covers what the product does today. Future features invite the question of why they're not live.
The founder's backstory. Powerful in a founder profile or podcast. In a 75-second explainer it delays the payoff.
Every feature. One action. The rest belong in separate short clips.
Acronyms without a reason. If a term needs explaining, cut it rather than explain it.

The cut list tends to shrink a first draft by a third. That's normal.

When doesn't this structure apply?

A few cases where I'd change the approach.

Developer-facing infrastructure. If your buyer is an engineer, a narrative explainer can feel patronising. A 90-second screen-recorded demo of a real integration, with the code visible, does the job better. Keep the problem shot, swap the story for the terminal.

Crypto-native audiences. If the video lives in a DeFi community that already knows what a vault or a bridge is, you can use their vocabulary and spend more time on what's different about you.

Pre-product. If there's nothing to show in beat 3, don't make an explainer. Make a founder video about the problem instead, and save the explainer until the product can carry its own middle.

There's also a cost question. A polished animated explainer takes real budget and goes stale when the interface changes. For early teams, a clean screen recording with a good voiceover is usually the better first version. My broader take on short video for startups is in YouTube and short video for B2B startups, and the one-sentence story your script hangs on is the same one I build first in any Web3 PR campaign.

FAQ

Should a crypto explainer video have a voiceover or on-screen text?

Use both, but design for sound-off first. Plenty of people scroll feeds with the sound off, so key lines should appear as captions or on-screen text. The voiceover then adds warmth for the people who do listen.

How many versions of the explainer should I make?

Make one master cut at 75 seconds and a short cut of 15 to 20 seconds using only beats 1, 2 and 5. The short cut works for paid social and X, and the master lives on your homepage and in pitch emails.

Can I use the explainer video in a press pitch?

Yes, as a link, never as an attachment. A short explainer helps a reporter understand the product quickly, but it doesn't replace a clear one-paragraph pitch. Treat it as supporting material.

How often should I update the explainer?

Whenever the interface shown in beat 3 changes noticeably, or when your proof point in beat 4 is out of date. An explainer showing old screens quietly tells viewers the product has moved on without them.

If your explainer opens with your logo and you suspect that's the problem, book a 30-minute script review and bring the draft.

I'm Shilika Jain, a fractional PR lead for Web3 and AI founders. More playbooks at shilikajain.com.

Conference Content Strategy: Capture Once, Publish for Six Weeks

Shilika — Thu, 05 Nov 2026 05:13:06 +0000

A good conference content strategy decides what you'll publish in the six weeks after the event before you capture anything during it. You plan a short list of content slots, capture against that list on the ground with a phone and a cheap microphone, log everything each night, and release it on a schedule long after the event hashtag has gone quiet. The capture takes a few hours across the week. The value comes from the weeks of posts it feeds.

Most teams get this the other way round. They film everything, post a highlight reel on the last day, and the rest of the footage dies in a camera roll.

Why does most conference content disappear after a week?

Because it's published at the worst moment. During the event, every team in the building is posting the same stage photos and the same "great week" thread. Your audience is either at the event, too busy to read, or at home, tired of seeing it.

Once the week ends, the feed goes quiet. That quiet stretch is when a sharp clip of your founder making one strong point, or a short interview with a partner, actually gets watched. The capture happens in the week. The publishing should mostly happen after it.

There's a second reason. Teams capture moments, not material. A crowd shot or a booth selfie is a moment. A founder answering a question in forty seconds, with clean audio, is material you can cut, quote, transcribe and repost for weeks.

What should you plan before the event?

Start with the content slots, not the camera. I write a simple list of what the team will publish after the event, then work backwards to what needs capturing.

A typical slot list for a founder-led startup looks like this:

One long-form recap post from the founder, built on two or three opinions the week confirmed or changed.
Three to five short clips of the founder answering a single question each.
One or two interviews with partners, customers or ecosystem leads.
A cut of the founder's talk or panel, if there is one.
One written piece, a newsletter issue or a blog post, that turns the week's conversations into a point of view.
A handful of photos that show real conversations, for use across posts for months.

Then decide who captures. On a small team, the person in the content seat does it, and the founder knows which questions they'll be asked on camera before they arrive. If you're still deciding who travels, my piece on who to send to a conference treats content capture as a proper seat with its own target.

Check one thing in advance: each organiser's rules on filming. Many events record main-stage talks themselves, so ask whether you can get the file of your founder's session instead of filming from the crowd. Ask partners and interviewees for consent before recording, and tell them where it will be published.

What can one person capture with a phone?

More than enough. You don't need a crew. You need clean audio, steady framing and a plan.

The kit I'd pack:

A phone with plenty of free storage
A clip-on wireless microphone, because conference halls are loud and bad audio ruins everything else
A small tripod or grip
A power bank
A notes app or a shared doc for the nightly log

Most of what you capture falls into four kinds of asset:

Single-question clips. Ask the founder one question, get one answer of under a minute. Shoot somewhere quiet, a corridor or outside the hall.
Short interviews. Five to ten minutes with a partner or guest, three or four questions, agreed beforehand.
Talk footage. From the organiser if possible. If not, film from a steady position with the mic on the speaker or close to the PA.
Written notes. The quotes, ideas and disagreements you heard in meetings and hallways, written down within the hour while you remember them.

I've done the interview part from the journalist side. In 2024 I spent TOKEN2049 and Solana's conference doing on-ground interviews.

I’d be onground @token2049 & @SolanaConf doing interviews for @cryptodailyuk!

Shilika Jain on X, 27 Aug 2024

Doing interviews on the ground teaches you quickly how little time each one takes when the questions are ready, and how much you lose when they aren't. A guest gives you ten good minutes between meetings. If you spend five of them working out what to ask, you get five.

How do you capture talks and interviews without a crew?

Treat each interview like a short booking, not a chat. Agree the slot, send three or four questions in advance, and tell the guest how long it'll take. Book them before the event, the same way you'd book a press or partner meeting.

On the day, keep a simple rhythm:

Mic the guest, test ten seconds of audio, play it back.
Ask the questions in the order you sent them.
End with one open question, such as what surprised them this week. The best clip often comes from that.
Take one still photo of the guest, with permission, for the post.
Write down the strongest line and its rough timestamp before you leave the spot.

Interviews also keep working long after the week they're recorded in. That autumn I was still publishing and sharing interviews, including one about zkPass.

was great to interview @Joshua_zkp about @zkPass ✨

Shilika Jain on X, 11 Oct 2024

An interview isn't news, so it doesn't expire with the event. What readers care about is whether the guest said something worth hearing.

For talks, the content is in the argument, not the stage. Capture or request the full recording, then cut the two or three moments where the speaker says something specific. If the talk itself is your main earned-media asset, the companion piece on turning a conference talk into a 30-day earned media campaign covers the press side. This post is about the content you own.

What is the nightly log, and why does it matter?

The nightly log is twenty minutes each evening where the person capturing writes down what they have. It's the habit that separates teams who publish for six weeks from teams who publish once.

Each entry has:

File name or location
Who's in it, and whether they consented to publishing
The single strongest line, written out
Which content slot it feeds
Anything that needs a check before it goes out, such as a claim, a figure or a partner's approval

By the end of the week you've got a list you can schedule from. Without it, you'll be scrubbing through hours of footage on the flight home trying to remember which clip had the good answer.

How do you turn one week into six weeks of posts?

This is the capture-to-calendar map I give founders. Each row is a kind of raw material. Each column is where it goes and when.

Raw capture	Week 1	Weeks 2 to 3	Weeks 4 to 6	Long-term use
Founder single-question clips	One clip with a short take on X and LinkedIn	One clip a week, each on a different theme	Remaining clips, tied to news when the topic comes up	Website, sales decks, pitch follow-ups
Partner or guest interviews	Short teaser	Full interview published, guest tags and reshares	Pull quotes as text posts	Partner pages, case studies
Talk recording	Best 60 seconds	Full talk with a written summary	Clip reused when the topic is in the news	Speaker reel for future applications
Written notes and quotes	Founder recap post	A newsletter or blog post built on one argument	An op-ed pitch if the argument holds up	Source material for future posts
Photos of real conversations	Recap post	Interview and clip thumbnails	Profile and article images	Press kit, website

Two rules keep this honest. First, one idea per post. A clip that tries to cover three points gets skipped. Second, each post should stand alone for someone who wasn't there. "What a week at the conference" means nothing to them. "Why wallet teams kept asking about recovery" does.

Platform choice matters too. If your founder is splitting time between channels, my piece on running LinkedIn and X as one founder content system covers which format suits which. For the video side, short video for B2B startups has the editing basics.

When the written material is strong but nobody has time to turn it into finished posts, that's the gap my content writing service fills: the notes come in raw, and the essays and posts go out on schedule.

When is a content capture plan not worth it?

It's a cost, so treat it like one.

If nobody on the team will publish afterwards, don't capture. Footage without a publisher is just storage.
If your founder isn't comfortable on camera yet, start with written notes and a recap post. A stiff clip does more harm than no clip.
If you're there purely for closed-door meetings, filming can make people guarded. Keep the camera away from those conversations and capture only public moments and agreed interviews.
If the event is small and private, ask the host before filming anything. Some of the best rooms are good precisely because nobody is recording.

The trade-off is time. Capture takes the content seat away from meetings for a few hours each day. If that seat also carries a sales target, decide in advance which one wins.

FAQ

Do we need a videographer for conference content?

Not for most startups. A phone, a clip-on microphone and a plan cover single-question clips, interviews and photos. A videographer starts to make sense for a hosted event you want filmed properly, or if you're producing a polished brand film, and then it's worth budgeting as its own line.

Should we post during the conference or after?

Do both, but weight it towards after. Post a little during the week so people know you're there and can find you. Hold most of the strong material for the weeks after, when the feed is quieter and your clips aren't competing with every other team's recap.

How long should conference clips be?

Short enough to make one point. For single-question clips that usually means under a minute. Full interviews and talk recordings can be longer on YouTube or your site, with short cuts on social platforms that link back to them.

Can we reuse a partner's interview on our own channels?

Only with their consent, and ideally in writing before you record. Tell them where it will appear and whether you'll cut it into clips. Send them the finished pieces before publishing so they can reshare, which is also how the interview reaches their audience as well as yours.

Want a capture plan built around your next event? Book a 30-minute session and leave with a slot list and a six-week release map.

I'm Shilika Jain, a fractional PR lead for Web3 and AI founders. More playbooks at shilikajain.com.

Off the Record vs On Background: Meaning for Crypto Founders

Shilika — Thu, 05 Nov 2026 05:12:57 +0000

"On background" means the reporter can use what you say but won't name you; they'll attribute it to a description such as "a person familiar with the matter." "Off the record" means the reporter agrees not to publish what you say at all, and usually not to use it unless they confirm it elsewhere. Both only apply if the reporter agrees before you speak, and definitions vary between newsrooms, so ask how their outlet uses each term. For most crypto founders, the better default is to stay on the record.

What do the attribution terms actually mean?

There are four levels you'll hear about. Here's the plain-English version, with the caveat that individual outlets and reporters sometimes draw the lines a little differently.

Term	Reporter can use the information	Reporter can quote your words	How you are identified
On the record	Yes	Yes	By name and title
On background	Yes	Often yes, as an unnamed quote	A description, e.g. "a person close to the project" or "a protocol executive"
Deep background	Yes, as the reporter's own understanding	No	Not at all
Off the record	No, unless they confirm it independently elsewhere	No	Not at all

The thing founders get wrong most often is treating "off the record" as a magic phrase that protects anything they say. It isn't. It's an agreement between two people, and like any agreement, it needs both of them to say yes, at the start, with the same understanding of what it means.

An embargo is a different animal. It controls when something can be published, not who said it. If you're mixing the two up, my embargo, exclusive and briefing decision framework separates them.

How do you agree on attribution before you speak?

You state the terms, the reporter accepts them, and only then do you say the sensitive thing. Announcing "that was off the record" after you've said it is a request, not a rule. A reporter may honour it as a courtesy. They don't have to.

I give founders a short script I call the four-line ground rules. It takes thirty seconds and removes most of the risk:

Name the term. "Before I answer, can we go on background for this part?"
Define it out loud. "By background I mean you can use the information, attributed to someone familiar with the project, but not my name."
Get an explicit yes. Wait for the reporter to agree. A nod on video or "sure" in a call counts; silence doesn't.
Mark the exit. "OK, back on the record now." Then say it again in writing afterwards if the conversation mixed terms.

If the reporter says no, you have your answer. Either say it on the record or don't say it.

Two practical rules sit around the script. Agree terms per topic, not per call; it's normal for most of an interview to be on record with one short background section. And if a PR person is on the call, they should confirm the terms in a follow-up email that same day, so nobody is relying on memory.

Why should founders mostly stay on the record?

Because your name is the asset. A founder interview exists to build a public, attributable voice: quotes people can search, link and cite. Every sentence said on background is a sentence that builds the reporter's story without building your reputation.

There's a trust argument too. Crypto has spent years in a credibility hole, and anonymous sourcing from projects doesn't help. Reputation in this industry is built in public, one attributable interaction at a time, and that's true for PR people as much as founders.

PR is the launchpad to build authentic & genuine reputation

Shilika Jain on X, 11 Jul 2025

A reputation can't be built from quotes nobody can attach to you. The founders who show up in the media by name are the ones who earn credibility from it. Reporters notice the difference between a source who stands behind their words and one who only talks in the dark.

On-record also keeps you consistent. Everything you say publicly has to agree with what your team, your docs and your community channels say. That's hard enough without a parallel set of unattributed comments floating around.

comms is so underrated in web3 but that's everything! how you communicate w your community, stakeholders, customers, retail users, airdrop hunters, journos is literally everything!

Shilika Jain on X, 3 Jun 2025

When is background the right call?

There are real situations where background helps both you and the reporter. The test I use is the record-default check: say it on record unless one of these is true.

Context, not claims. You're explaining how something works (a bridge design, a market-maker loan structure, why a vesting schedule looks the way it does) so the reporter gets the technical picture right. The facts matter; your name on the explanation doesn't.
Pointing to public evidence. You're telling a reporter where to look (an on-chain transaction, a governance forum thread, a filing) so they can verify it themselves.
Industry sensitivity. You're describing a pattern across projects you work with and naming yourself would identify a partner who hasn't consented.
Live legal or security process. Something is under investigation and you can share direction but not detail. Get your lawyer's view first; background isn't a legal shield.

And when it isn't the right call:

To criticise a competitor without owning it.
To float a claim about your own project you're not willing to defend.
To test a story you haven't finalised internally.
Anything about token price, listings or partnerships you can't confirm.

If any item on that second list is the reason you want background, the honest answer is not to say it.

What can go wrong with off-the-record conversations?

More than founders expect. A few failure modes I plan around:

Mismatched definitions. You think off the record means "never use this." The reporter thinks it means "don't quote me but I can chase it." Both are common understandings. That's why step two of the script matters.
Independent confirmation. Even a scrupulous reporter can publish information you shared off the record if a second source confirms it. You told them where to dig.
Group settings. A dinner, a panel green room or a Telegram group with several people is not an off-the-record setting just because you say so. Everyone in the room would need to agree, and they won't all be bound the same way.
Events. At conferences, reporters are working. A casual chat at a side event can end up in a story if no terms were set. I tell founders at events to assume they're on record by default unless they've agreed otherwise with that reporter.
The burned source problem. If you use off-the-record too often, reporters learn that you're useful only in private. That's not the relationship you want.

How do these terms play out in a critical story?

Pressure makes founders reach for background most. A reporter is writing something unflattering, and the instinct is to explain your side without being quoted.

Sometimes that's sensible: walking a reporter through the facts on background can correct a wrong assumption before it hits print. But if the story will run with or without you, an on-record statement is usually stronger, because it puts your side in the article under your name. A source "familiar with the project" disputing a claim reads weakly next to named critics.

Responding to a critical story is its own discipline, covered in my piece on how to respond to a journalist request for comment. The short version for this question: decide your on-record position first, then decide whether any background context would genuinely help accuracy.

A worked example: one interview, three terms

This is what a clean version looks like. A founder is doing a 30-minute call with a trade reporter about a protocol upgrade.

Minute	What's discussed	Terms	Why
0 to 20	The upgrade, the reasoning, the roadmap	On the record	This is the story; the founder wants to be quoted
20 to 25	Why a rival design was rejected, including a partner's security concern	On background, agreed before speaking	Useful context for accuracy; naming it would expose a partner
25 to 30	Back to adoption plans and next milestones	On the record, marked out loud	Ends the call on attributable quotes

After the call, the PR lead emails a one-line recap: "Thanks for today. To confirm, minutes 20 to 25 on the rival design were on background; everything else on the record." No drama, no ambiguity.

That's how I prepare founders in founder profiling work: the attribution plan is part of the interview brief, not an afterthought.

When this advice doesn't fully apply

If you're a whistleblower, a source inside a company that isn't yours, or someone facing genuine safety risks, the calculation is different and anonymity can be essential. This post is written for founders and executives speaking about their own projects, where being named is usually the point.

Norms also differ across countries and outlets. Some newsrooms have stricter rules about anonymous sourcing than others, and some avoid unnamed quotes entirely. When in doubt, ask the reporter how their editor treats background material before you agree to anything.

For more on preparing for interviews in general, see the first media interview checklist, and if you want to become the founder reporters call on record, read how to become a source for crypto journalists.

FAQ

Can I go off the record by email?

You can ask, but write the terms at the top of the email and wait for the reporter to agree before sending the sensitive part. Never put something in writing that you'd regret seeing quoted, because emails get forwarded and saved. For anything truly sensitive, a call after agreed terms is safer.

Does "no comment" count as off the record?

No. "No comment" is an on-record response, and it's often printed as "declined to comment." If you have nothing to add, a short on-record line explaining why you can't comment usually reads better.

Can a reporter name me if I said something off the record by mistake?

If you didn't agree terms first, the reporter is generally within their rights to treat it as on the record. Many will be fair if you ask quickly and the slip was minor, but that's goodwill, not an obligation.

Should my PR person stay on the call?

For first interviews and anything sensitive, yes. They can keep track of which parts were agreed as background and confirm terms in writing afterwards, which protects both you and the reporter.

Have a sensitive interview coming up and unsure what to say on record? Get in touch and I'll help you set the ground rules before you dial in.

I'm Shilika Jain, a fractional PR lead for Web3 and AI founders. More playbooks at shilikajain.com.

Story Bank Template: Collect Stories That Feed PR and Content

Shilika — Wed, 04 Nov 2026 05:13:38 +0000

A story bank is a shared spreadsheet or database where your team logs real moments as they happen (a user's comment, a bug that taught you something, a partner's surprising use case, a number that moved) so that PR, content and KOL briefs never start from a blank page. A useful story bank template has about ten columns: the story in one line, the source, the proof, permission status, the narrative theme it supports, a ready score, and where it's been used. The hard part isn't the template. It's the weekly habit of filling it.

What is a story bank, and why does a startup need one?

Most teams have plenty of stories. They're just scattered across Telegram chats, call notes, Discord threads and the founder's memory, and they disappear within a week.

Then a reporter asks for an example of a real user, or a KOL needs a hook, or the founder has to write a post on Thursday, and everyone says the same thing: "We had a great one a few months ago, but I can't remember the details."

A story bank fixes that. It's a single place where raw material goes the moment it appears, with enough context that someone else can use it later. It's the supply side of every narrative you're trying to build. If you've chosen what your company stands for, the bank is where the evidence for it lives.

It also matters more now than it did a few years ago.

now brands need folks that could write old-school original content without using AI

Shilika Jain on X, 19 Jun 2026

AI can produce a decent paragraph about almost anything. It can't produce the thing your user said on a call last Tuesday. Original, first-hand stories are what separate your content from everyone else's, and a story bank is how you keep a stock of them.

The story bank template: columns that earn their place

I keep this in a spreadsheet for most teams. Notion or Airtable works too, but a sheet is what people actually open. I've been a fan of simple trackers for a long time:

structured one pager/briefs for external sharing

google sheet trackers

Shilika Jain on X, 20 May 2024

These are the columns, and why each one exists.

Column	What goes in it	Why it matters
Date logged	When it was added	Freshness: old stories need re-checking
The story in one line	"A DAO treasurer moved funds off an exchange after trying our recovery flow"	Scannable; if you can't write one line, it's not a story yet
Source type	Team, user, partner, data, market	Shows if you're over-relying on one source
Who to ask	Name of the person who knows the detail	So anyone can follow up
Proof	Link, screenshot, on-chain tx, call recording, metric	Reporters and KOLs need evidence, not anecdotes
Permission	Public, ask first, internal only	Stops someone posting a user story without consent
Theme	Which narrative pillar it supports	Keeps the bank tied to your story, not random
Ready score	0 to 8 (see below)	Tells you what can be used this week
Best route	Press, founder post, KOL brief, website, sales	Speeds up the hand-off
Used in	Links to where it's been published	Prevents the same story being overused

Ten columns. If you add more, people stop filling it in. If you drop permission or proof, sooner or later something embarrassing goes out.

How do you get stories out of your team?

Asking "does anyone have a good story?" in a group chat gets silence. Specific prompts get answers. I give each source its own prompts, phrased as asks, and run them weekly.

Team (engineering, product, support)

Tell me what broke this week and what fixing it taught you.
Share a user question that surprised you.
Name something you decided not to build, and the reason.

Users and community

Paste one thing someone said in Discord or Telegram that the founder should see.
Flag anyone who used the product in a way we didn't expect.

Partners

Note what their users are doing with the integration that we didn't plan for.
Capture anything a partner's team said on a call that explains why they chose us.

Data

Pick the number that moved this week and say whether we know why.
Spot any pattern in on-chain activity that tells a story by itself.

Market

List the news this week we have a real opinion on.
Write down one thing the category gets wrong that we see every day.

The single best habit I've found is a 20-minute "story harvest" at the end of each week. One person runs it, asks the prompts, and logs anything with a pulse. Rough is fine. Someone can tidy it later.

The ready score: which stories can you use this week?

Not every entry is ready to go out. I score each story on four checks, 0 to 2 each, for a total out of 8.

Proof (0 to 2). 0 is "someone remembers it". 1 is a screenshot or internal note. 2 is a public link, on-chain record or a number you can source.
Permission (0 to 2). 0 is unknown. 1 is "probably fine, ask first". 2 is written approval or fully public.
Freshness (0 to 2). 0 is older than six months. 1 is one to six months. 2 is this month.
Fit (0 to 2). 0 is interesting but off-message. 1 supports a theme loosely. 2 is direct evidence for one of your core themes.

A score of 7 or 8 can go into a pitch or a KOL brief now. A 4 to 6 needs one fix, usually permission or proof, so assign that fix to someone. Under 4, keep it in the bank as background, but don't build anything on it yet.

The score stops the most common mistake I see: a founder telling a great user story to a reporter, the reporter asking to speak to the user, and nobody having asked the user first.

Where each story should go

The routing column turns a pile of stories into a content plan. Different kinds of stories suit different channels.

Story type	Best first route	Second life
A number with a clear cause	Press pitch or data post	Founder post on X, website proof point
A user's own words	Website, sales deck, founder post	Case study once permission is signed
A partner's unexpected use case	Joint content with the partner	Pitch angle for trade media
A team decision or failure	Founder post or behind-the-scenes content	Podcast talking point
An opinion on a news event	Reactive commentary to reporters	Op-ed if it holds up
A regional or local angle	Regional media pitch	Localised KOL brief

For opinions on breaking news, speed matters, so these should sit at the top of the bank with an owner who can respond the same day. My reactive commentary guide covers how that works in practice. For team decisions and process, the Glass Wall Sort in my behind-the-scenes content post helps decide what's safe to share.

Stories don't need a bank to exist, but having them ready changes how fast you can move. When the global Microsoft outage hit in 2024, Fluence had a clear argument about why decentralised compute matters, and that turned into an e27 piece tying the outage to the need for DePIN. Commissioned research did something similar for Elastos: a finding that 67% of Nigerians trust Bitcoin for life savings over traditional methods earned its own Cryptonews headline. A number, a place and a clear angle. Those are exactly the entries a story bank should be full of.

How do KOL briefs and press pitches use the bank?

This is where the bank pays for itself.

A KOL brief built from the bank has real hooks: a user quote, a number with context, a decision the team made. Creators can work with that. A brief built from the website copy gives every creator the same three sentences, and every post sounds the same. My KOL brief template has a slot for exactly this kind of story.

A press pitch built from the bank can answer the reporter's follow-up questions before they ask: who the user is, whether they'll talk, where the number comes from. If those are already in the proof and permission columns, you reply in minutes instead of days.

I treat the media side the same way. Years ago I set out to build my own publications database:

gonna build a massive database of publications covering topics related to web3, blockchain, and crypto

Shilika Jain on X, 10 Jul 2023

A media database tells you where a story could go. A story bank tells you what you have to send. You need both, and most teams only have half of one.

It also keeps your channels telling one story. If everyone pulls from the same bank, the founder, the website and the KOLs are citing the same evidence. The sibling post on narrative consistency across channels goes further on that.

When a story bank isn't worth it

Be honest about whether you'll maintain it.

Teams of two or three. If everyone sits in the same calls, a shared notes doc might be enough. Move to a proper bank once stories start getting lost.
No one owns it. A bank without an owner becomes a graveyard of half-entries within a month. If you can't name the owner, don't start.
No narrative yet. If you haven't decided what you stand for, the Theme column will be empty and the bank becomes a random collection. Pick your themes first.
Highly confidential products. Some enterprise or security work can't be discussed at all. The bank can still hold market opinions and team stories, but expect it to be thinner.

The ongoing cost is about half an hour a week for the harvest, plus whoever tidies the entries. In return, every piece of content starts with real material.

Setting up the template and running the first few harvests with a team is part of my content writing work, until it runs without me.

FAQ

Should the story bank be visible to the whole company?

Yes, read access for everyone and edit access for a few. People contribute more when they can see their stories being used. Keep a separate private tab for anything marked internal only, like sensitive user conversations or unannounced partner details.

How many stories should a story bank have?

There's no target number, but a healthy bank gets a handful of new entries each week and has several stories scoring 7 or 8 at any time. If you go a fortnight without a new entry, the harvest has stopped, not the stories.

How do I get permission to use a user's story?

Ask them directly, explain where it will appear, and get the answer in writing, even if it's just a message. Offer to anonymise it if they're hesitant. For press, check whether they're willing to speak to a reporter, which is a separate and bigger ask.

When should a story be retired from the bank?

When the proof goes stale (the number has changed), when the person involved has left or withdrawn permission, or when it's been used so often your audience will recognise it. Move retired stories to an archive tab rather than deleting them; they're still useful background.

Sitting on stories your team never writes down? Grab 30 minutes with me and we'll set up your story bank and the first harvest together.

I'm Shilika Jain, a fractional PR lead for Web3 and AI founders. More playbooks at shilikajain.com.

Format	Who writes it	How it's usually marked	What it can do	What it can't do
Sponsored post or partner content	You, sometimes edited by the outlet's commercial team	"Sponsored", "partner content", "paid post" or similar	Put a long explainer on a known domain	Signal that a journalist chose you
Press release section	You, word for word	"Press release", often a separate section	Create a public record of an announcement	Reach the outlet's regular readers in any depth
Newsletter sponsor slot	You, in a fixed format	"Sponsored" or "presented by"	Get a short message into a curated inbox	Build credibility on its own
Paid podcast or video segment	Host plus your talking points	Varies a lot, sometimes weakly	Give a founder airtime	Count as an independent interview

What you want	Better fit	Why
Credibility with investors and reporters	Earned coverage	Only an editor's choice signals independent judgement
A funding or launch story in tier-1 press	Earned, with an embargoed pitch	Paying first uses up the news
A detailed explainer on a niche, well-matched site	Sponsored, clearly labelled	You control depth and accuracy
A routine update that just needs to exist publicly	Your own blog, optionally a wire	Neither reporters nor paid readers care much
Reach into a specific community quickly	Newsletter sponsorship or KOLs	Faster and better targeted than a sponsored article

How to Use AI in a PR Workflow: Draft, Assist or Keep Human

Shilika — Tue, 03 Nov 2026 05:13:14 +0000

Use AI in a PR workflow by sorting every task into one of three lanes. In the draft lane, AI produces the first version and a person edits it. In the assist lane, AI speeds up a person's research or thinking but never produces the final output. In the human lane, AI stays out. Every AI output then gets one named verification step before it touches a reporter, a founder or the public.

That's the whole method. The rest of this post is the map I use to decide which lane each task belongs in, stage by stage, and the specific check I run on each output.

What does a PR workflow actually involve?

People who haven't run PR tend to picture one task: asking a publication to write about you. I've pushed back on that for years.

what it actually is - creating a storyline, working on the messaging & narrative, pitching to targeted journalists, reaching out to them with well-curated pitches, following up & more!

Shilika Jain on X, 13 Apr 2023

Broken into stages, a typical campaign for a launch or a funding announcement runs roughly like this:

Research and monitoring
Narrative and angle
Media list building
Pitch writing
Outreach and follow-up
Interview and spokesperson prep
Releases, op-eds and owned content
Localisation for other markets
Coverage reporting
Repurposing coverage into owned and search-friendly content

Each stage has a different mix of production work and judgment. That mix decides the lane.

The three-lane map: draft, assist, human

This is how I'd sort each stage today. The verification column matters most, because "AI did it" is never a finished state.

Stage	Lane	What AI does	Verification before it's used
Research and monitoring	Assist	Summarises a reporter's recent articles, flags mentions	Open the original articles; never pitch off a summary alone
Narrative and angle	Human	Lists options you then reject	Founder and comms lead agree on one angle in writing
Media list building	Assist	Formats, de-duplicates, suggests beats	Check each reporter's last three bylines yourself
Pitch writing	Draft (lightly)	Subject-line variants, tightening	Read it aloud; remove anything you wouldn't say on a call
Outreach and follow-up	Human	Nothing, or reminders only	Personal send, personal follow-up
Interview prep	Assist	Generates likely hostile questions	Spokesperson rehearses with a person
Releases and op-eds	Draft	Structure and first pass from approved facts	Fact sheet check, every number and name traced to a source
Localisation	Assist	Rough translation for internal review	Native-speaker edit plus local outlet fit
Coverage reporting	Draft	Compiles links, summaries, tables	Click every link, confirm earned vs paid
Repurposing	Draft	Turns one interview into posts and FAQs	Founder approves anything in their voice

A few of these need more explanation than a table cell allows.

Where AI drafts well: releases, reports and repurposing

The draft lane works when the inputs are already fixed. A press release built from an approved fact sheet, a coverage report built from a list of links, a LinkedIn post built from a transcript the founder already said out loud. The thinking happened earlier, so the model is only arranging it.

My rule for the draft lane: AI never gets to add a fact. It can reorder, shorten and restructure. If a number, name, date or quote appears in the draft that wasn't in the input, it comes out until someone can trace it. Invented quotes and confident wrong figures are the main way AI-drafted releases go wrong, and I go through those risks properly in the risks of using AI to write press releases.

Repurposing is where the time savings are largest. One founder interview can become a thread, a newsletter section, an FAQ block for your site and a short op-ed outline. That last step also helps with AI search, since answers that live on your own pages in clear question-and-answer form are easier for assistants to pick up. I cover that side in why PR drives AI search citations.

Where AI should only assist: research, lists and localisation

The assist lane is for tasks where AI speeds you up but a wrong output is expensive.

Research. Asking a model to summarise a reporter's recent work is a fine starting point. Pitching from that summary is not. Models blur reporters together, miss beat changes, and occasionally attribute articles to the wrong byline. Always open the originals before you write a word of the pitch.

Media lists. A good list is a set of judgments about who covers what this quarter, not a spreadsheet. AI can clean the spreadsheet. I've written about building one properly in how to build a media list, and none of that process is skippable because a tool exists.

Localisation. This one I feel strongly about because so much of my work is APAC. Machine translation gives you a rough internal draft. It doesn't tell you that Korean reach often runs through Naver-syndicated mainstream outlets and Telegram-driven news apps like CoinNess, which is something we learned placing coverage across Korea, and it doesn't rework the angle for what a local editor cares about. Translating an English release and hoping is the failure mode, with or without AI. When Tangem's self-custody survey was turned into a data story, one stat travelled across English, Korean, Japanese and Dutch editions without a separate pitch for each market. That travelled because the stat was strong, not because the translation was fast.

What stays human, and why it's the expensive part

The human lane is short but it's where campaigns succeed or fail.

edit pitches

eat lunch

pitch pitches

eat dinner

follow up on pitches

Shilika Jain on X, 23 May 2024

I posted that as a joke about the life of a PR manager, but it's accurate about where the hours go. Editing, pitching and following up are relationship work. A reporter can tell when the follow-up came from a sequence tool, and a templated nudge on a pitch they already ignored makes the next pitch less likely to be opened.

The narrative and angle stage also stays human. AI is useful for generating options so you can see what you're rejecting. The choice itself depends on what reporters are tired of, what competitors already said and what the founder can credibly defend in an interview. I'd never let a tool pick the angle for a funding announcement.

The verification rule set I use on every AI output

Each lane has its own checks in the table, but five rules apply to everything that comes out of a model:

Trace every fact. Any number, name, date, title or quote must point to a source document a person has opened.
Read it aloud. If a sentence sounds like nobody in particular wrote it, a reporter will hear that too.
Check the person, not the summary. Reporter details, beats and recent articles come from the outlet's site, not the model's memory.
Mark earned vs paid. In coverage reports, every link gets checked so sponsored posts aren't counted as earned coverage.
Get the voice owner's sign-off. Anything published under a founder's name gets the founder's approval, line by line if needed.

How do you roll AI into an existing PR workflow?

If your team isn't using AI yet, don't change everything at once. A sequence that works:

Week 1: Reporting. Start with coverage reports, the lowest-risk draft-lane task. You'll learn how the tool fails without any external exposure.
Week 2: Research summaries. Add reporter and topic summaries, with the rule that nobody pitches without opening the originals.
Week 3: Repurposing. Turn existing interviews and coverage into owned content, with founder sign-off.
Week 4: Release first drafts. Only from a locked fact sheet, with the trace-every-fact rule enforced.

Pitches and outreach stay out of the rollout entirely. Those are the steps where a bad AI output costs you a relationship rather than an hour.

When this map doesn't fit

Two situations change the lanes. If you're a solo founder with no comms person, you'll lean on the draft lane more heavily for pitches too, and that's reasonable as long as you rewrite them in your own words before sending. And if your company works in a regulated area, such as finance, health or anything with securities questions, move releases and op-eds from the draft lane to the assist lane, because every sentence will need legal review anyway and a model's confident phrasing creates more redlines, not fewer.

The map also assumes someone senior is reading the output. AI in a PR workflow saves real hours, but only for teams that already know what good looks like.

FAQ

Which AI tools are best for PR work?

General assistants like ChatGPT and Claude handle most of the drafting and summarising described here. The tool matters less than the lane rules: a better model still shouldn't pick your angle or send your follow-ups. For a wider view of startup tooling, see my notes on the AI marketing tool stack for startups.

Should I tell reporters I used AI?

You don't need to announce that AI helped tighten a pitch. What matters is that every fact is accurate and the pitch reads as written by someone who knows their beat. If AI generated a quote or a statistic, that's not a disclosure question, it's an accuracy problem to fix before sending.

How much time does AI actually save in PR?

It saves the most on reporting, repurposing and first drafts, and close to nothing on relationship work. Teams that try to save time on pitching and follow-up usually lose it back in lower response rates. Measure it on your own workflow for a month before deciding.

Can AI build a media list for me?

It can suggest names and clean up a sheet. It can't confirm who's actually covering your topic this quarter or who has moved outlets. Treat its list as a starting draft and verify every reporter against their recent bylines.

Want a second pair of eyes on where AI fits in your comms setup? Book a 30-minute PR workflow teardown, or see how I run Web3 PR campaigns end to end.

I'm Shilika Jain, a fractional PR lead for Web3 and AI founders. More playbooks at shilikajain.com.

How Journalists Find Sources on Twitter, and How to Be One

Shilika — Tue, 03 Nov 2026 05:13:07 +0000

Journalists find sources on X in four main ways: searching keywords when news breaks, watching curated lists of people who post knowledgeably on their beat, noticing who replies with substance under their own posts, and following the trail of who other trusted sources talk to. Once a name surfaces, they check the profile for credibility and send a DM or an email. If you post specific takes in one niche, state your expertise in your bio, keep your DMs open and answer fast, you become someone they can find.

I've worked on media relations in crypto for years, and the shift I've watched is that X has become the place reporters go first when they need a reaction quickly. Founders and operators who understand how that sourcing works get quoted. Those who don't keep sending press releases into inboxes instead.

How do crypto journalists use X to find sources?

From the outside it looks random. It isn't. A reporter on deadline has maybe an hour to find two or three credible voices on a story, and X is the fastest place to look. In my experience these are the main routes a name reaches them:

Keyword search during breaking news. An exploit, a regulatory move, a large token release. The reporter searches the protocol name or the topic and reads who's saying something specific.
Beat lists. Many reporters keep private lists of people who post well on their area: DeFi, L2s, stablecoins, AI agents. Getting onto those lists is quiet and slow, and it's where repeat quotes come from.
Their own replies. When a reporter posts a question or a story, the people who reply with something useful get noticed. Reporters read their mentions.
The trust trail. If a source they already rate is debating you, or reposts you, you inherit a little of that trust.
Direct callouts. Some reporters post "looking to speak to anyone who..." requests. Answering those quickly and to the point is free coverage most people ignore.

None of these involve a pitch. That's the appeal for the reporter, and the opportunity for you.

What makes an X profile look like a credible source?

A reporter who finds your post will click your profile next. They're asking one question: can I put this person's name and title in a story without embarrassing myself. The profile has to answer it in seconds.

I think of it as a five-layer source-readiness stack. Each layer depends on the one before.

Layer	What the reporter needs	What it looks like on your profile
Findable	To surface you in a search or list	Your niche's keywords in your bio and recent posts
Credible	To trust your title and context	Role, company and one line on what you know about
Quotable	A sentence they can lift	Recent posts with clear, specific claims, not vibes
Reachable	To contact you on deadline	Open DMs, and an email or contact route in your bio
Reliable	To come back next time	A track record of answering fast and being accurate

Most founders stop at "credible". They have a good bio and nothing quotable underneath it, or great takes and closed DMs. If you're working on the bio layer specifically, my guide to writing a founder bio for journalists and conference programmers goes through it line by line.

What kind of post gets you quoted?

A quotable post makes one clear claim, backs it with something specific, and is short enough to lift whole. It sounds like a person with an opinion, not a press release.

Compare two reactions to the same piece of news:

Unquotable: "Huge news for the ecosystem today! Excited to see where this goes. 🚀"
Quotable: "This is the third bridge exploit this year that started with a compromised signer, not a code bug. Audits won't fix key management."

The second gives a reporter a fact, a pattern and a position in two sentences. It might be wrong, and that's fine. Reporters quote positions, then balance them with other views.

A few habits help:

Post within a few hours of news in your niche. After a day, most stories are written.
Lead with the claim. Don't warm up.
Add one detail only you'd know from doing the work.
Avoid talking up your own product in the take. A reporter can't use a sales line.
Write single, standalone posts for reactions. Threads are better for explainers, which I cover in threads vs single posts.

I've seen the same principle work in pitched commentary. Magna, a token management platform, got into Decrypt with commentary on Worldcoin's tokenomics at launch without having any news of its own. The quote worked because it was about something reporters were already covering, and it said something specific.

How fast should you respond when a journalist DMs you?

As fast as you can, ideally within the hour on a breaking story. If a reporter has DMed three people and you answer last, the story may already be filed. Answering a reporter's question promptly is the most basic part of media relations, and it's the part most founders get wrong, usually by waiting for a perfect answer instead of sending a good one now.

That applies to uncomfortable questions too. In November 2025 I wrote about a public back-and-forth between a crypto investigator and a project, and my view hasn't changed: if an investigative journalist keeps coming back to you with specific questions, answer sooner rather than later. Silence doesn't make the story go away. It just means your side isn't in it.

If you can't comment, say so quickly and say when you might be able to. "Not able to comment on this yet, happy to speak once it's announced" keeps the door open. Ignoring the DM shuts it.

What etiquette keeps journalists coming back?

Being quoted once is luck. Being quoted repeatedly is a relationship, and relationships run on a few rules.

Do:

Engage with their work when it's good, not only when it mentions you
Correct errors privately and politely, never with a public dunk
Keep your quote accurate to what you actually know
Say when a question is outside your expertise and suggest someone better
Respect embargoes and off-the-record agreements absolutely

Don't:

Ask to see the article before it runs, or to change your quote afterwards
Use a reporter's request as a chance to pitch your launch
Complain publicly about how you were quoted
DM a reporter every time you post something
Treat them as part of your marketing team

That last one is the root of most bad behaviour. Reporters work for their readers and their publication, a point I go into in journalists are not on your payroll.

Genuine engagement with reporters' work is underrated. I've asked my followers more than once to point me to the crypto journalism they admire.

whose your favvv crypto/web3 journalist?

Shilika Jain on X, 15 Nov 2022

It's a small habit, but it keeps me reading reporters as people doing hard work rather than as distribution. And those relationships have carried my own career as much as any placement has.

i wouldn't have survived in the industry so long w/o the support of the best journos, podcasters & fellow peers 🙏🏻

Shilika Jain on X, 11 Jul 2025

Reporters notice who reads their work closely. Thoughtful replies to their stories, especially investigative ones, build the kind of familiarity that makes your DM get opened later.

When does being a source on X backfire?

When you comment on things you don't understand. Being quoted wrongly in a story about an exploit or a regulatory action is worse than not being quoted, because that story gets read, shared and screenshotted for years. Stick to your actual niche.

It also backfires when you're in the story. If your own company is facing a crisis, your personal X account is not the place to be a source, and your comments should go through whoever is handling communications. Reacting on X in the heat of it is how a manageable story becomes a bigger one.

And it isn't a substitute for a press strategy. Being findable on X brings you reactive quotes, which are valuable for building a profile but rarely carry your own announcements. For those you still need pitching, relationships and timing. The two work together: the reporter who quoted you in March is more likely to open your launch pitch in June. That combined approach is what I build with founders through founder profiling, and my piece on reactive commentary for Web3 and AI founders covers the pitching side.

FAQ

Do journalists actually DM sources on X, or do they use email?

Both. For breaking news, a DM is often faster, which is why open DMs matter. For longer features, reporters usually move to email or a call once they've found you. Make sure your bio gives them a route to either.

Should I tag a journalist in my post to get their attention?

Rarely. Tagging a reporter on your take about their story can feel like pressure, and doing it often gets you muted. A substantive reply under their own post is more natural and more likely to be read.

How do I get onto a reporter's private list of sources?

You can't ask your way onto it. You get there by posting consistently and specifically on one beat over months, and by being useful in public conversations the reporter follows. Being accurate and quick the first time they contact you matters most.

Can a company account be a source, or does it have to be a person?

Reporters almost always quote people, not brand accounts. A company account can share news, but the quotable voice should be a named founder or executive with their own profile. That's one reason founder profiles matter as much as the company's.

Want reporters to find you before your next announcement rather than after it? Tell me what you're building and I'll show you where your profile stands.

I'm Shilika Jain, a fractional PR lead for Web3 and AI founders. More playbooks at shilikajain.com.