DEV Community: Yanmiayn

How I build B2B lead packs for AI automation agencies (exact row format)

Yanmiayn — Tue, 12 May 2026 03:47:51 +0000

The problem with bulk lead databases

Most B2B lead databases sell you 10,000 rows of noise. 40% bounce. 80% don't match your ICP. Zero rows have a first line your team can actually use.

I build the opposite: 50-250 rows per pack, each with a public source URL, a one-line ICP fit note, and a personalized first sentence written specifically for that company.

Exact row format

Every row includes:

Company name + website (verified live)
Public source URL — the page where the contact info was found, no LinkedIn scrape
ICP fit note — one line on why this account matches your offer
Personalized first line — references a real verifiable detail about that company
Email field formatted for NeverBounce or ZeroBounce

Example first line: Saw Quay Dental asks new patients to complete a detailed form before visiting — exactly where intake automation saves front-desk time.

No generic openers. Every row is verifiable.

Who this is for

AI automation agencies selling into dental, HVAC, MSP, legal, or accounting practices
Founders starting cold email who want 50 clean rows before committing to Apollo or Clay
Sales teams who need a targeted list for a specific campaign

Pricing

50 rows: $49
150 rows: $99
250 rows/week ongoing retainer: $199/week

Delivered as Google Sheets plus CSV within 24-72 hours. If more than 20% of rows are unusable for the agreed ICP, I replace or refund.

Get a free 5-row preview

DM me your ICP in one sentence. I will send 5 rows back before any paid order so you can check fit and format.

More at yanmiayn.com.

A small checklist before shipping a Claude API workflow

Yanmiayn — Fri, 08 May 2026 23:36:44 +0000

Most Claude API demos work on day one. Production workflows usually fail around the edges.

Before I ship a Claude workflow for a SaaS app, I check these parts first:

1. Is the output structured?

If the next step parses prose, the workflow will break. I prefer JSON schema or another explicit contract so the app can consume the result safely.

2. What happens on retry?

Rate limits, partial failures, and timeout paths need a boring answer. The workflow should know when to retry, when to stop, and what to log.

3. Where do the keys live?

Customer API keys should stay in the customer's environment or secrets manager. They should not be sent over email, DMs, Loom, or GitHub issues.

4. Is there a human approval step?

Anything customer-facing should usually start with a draft + approval flow. This keeps the first version useful without pretending the model is perfect.

5. Can another developer operate it?

A small README, .env.example, and Loom handoff are part of the product. If nobody can run it after delivery, it is not really shipped.

I opened a focused Fiverr gig for this exact thing: one Claude API workflow, async delivery, customer keys/customer infra, fixed scope.

$150 audit / $750 build:
https://www.fiverr.com/miconnm/build-a-production-claude-api-workflow-for-your-saas-in-7-days

why your cold outreach feels like ai slop (and how to fix it in 10 minutes)

Yanmiayn — Fri, 08 May 2026 01:00:46 +0000

why your cold outreach feels like ai slop (and how to fix it in 10 minutes)

there's a 426-point HN thread today called "AI slop is killing online communities."

the top comment is basically: we can tell. we can always tell.

cold outreach has the same problem.

what people actually mean when they say "personalized"

most founders treat personalization as a mail-merge field. {{first_name}}. maybe {{company}}. done.

that's not personalization. that's a slot machine with the same message behind every pull.

real personalization is noticing something specific and saying it plainly. no preamble. no throat-clearing. no "i came across your work and was blown away."

just: what you saw, why it matters to you, what you want.

three sentences. that's the whole formula.

the pattern that actually reads as human in 2026

i've been collecting cold DMs and emails that got responses. not templates — actual messages that worked. the pattern:

1. a reference only a human who paid attention would make

not "i saw your tweet." more like: "noticed you switched from monthly to annual pricing last month — curious what changed."

that's one sentence. it signals you're real.

2. a specific ask with a specific number

"do you have 20 minutes" is vague. "tuesday 3pm ET, 20 min, i'll send a loom first so you can decide if it's worth your time" is a real offer.

3. no signature block with three logos and a legal disclaimer

just your name. maybe a link. that's it.

why this matters more now than 6 months ago

inboxes are flooded with AI-generated outreach. everyone's tools got better at generating volume. so the signal-to-noise ratio collapsed.

the people still getting responses are the ones who sound like they typed the message themselves and meant it.

this is a rare case where doing less — fewer messages, slower, more specific — actually wins.

what "control flow" has to do with outreach

there's another thread today: agents need control flow, not more prompts.

the same idea applies to human outreach sequences. most people treat follow-up as "send message 2 on day 3, message 3 on day 7." linear. no branching.

but real sequences branch:

did they open? did they click?
did they reply with a question vs a no?
did they visit your site after the first DM?

each of those is a different branch. a different next step.

most cold outreach fails not because the first message was bad but because the follow-up ignored every signal after it.

the actual 10-minute fix

pull your last 5 outreach messages. for each one, ask:

does this reference something that couldn't be auto-generated?
does the ask have a specific time, duration, or deliverable?
is there a branch in my follow-up based on what they actually did?

if you answer no to any of those, rewrite before sending more volume.

the goal isn't to sound less like AI. it's to sound more like someone who actually did the homework.

if you want the exact message frameworks i use for cold DMs and emails (including the 3-sentence opener, the follow-up branching map, and 9 subject lines that don't sound like subject lines), i put them in a kit:

cold outreach kit — $9

no upsell, no drip sequence after. just the doc.

yanmiayn.com — tools and playbooks for one-person businesses

your cold outreach isn't broken. your target list is.

Yanmiayn — Thu, 07 May 2026 01:01:05 +0000

your cold outreach isn't broken. your target list is.

spent 3 months fixing the wrong thing.

better subject lines. shorter emails. different cta. open rates went up a little. replies didn't move.

then i looked at who i was emailing. that's where the problem actually was.

what a bad target list looks like

"founders" pulled from a twitter list someone posted in 2023
people who follow the same newsletters as you (they've seen 1000 pitches)
"decision makers" from apollo with job titles that match but context that doesn't
anyone you found by searching "ceo" + your industry keyword on linkedin

these people are either burned out on outreach or just wrong for what you sell.

what a good target list looks like

signal: they just had a problem you solve

someone posted about struggling with brand visibility 2 weeks ago
a founder just raised a seed round (new pressure to show up professionally)
a consultant just went independent (no team, needs to punch above their weight)
a company just hired its first marketing person (brand questions incoming)

the message writes itself when the context is right.

the actual research process (no paid tools)

step 1: pick one trigger event

don't try to target 5 different situations. pick one. for me it was: consultants who just went independent in the last 60 days.

step 2: find them where they announce it

linkedin "i'm excited to share" + your niche keywords (free search, public posts)
twitter/x searches: "going indie" OR "just quit" + consultant/founder/agency
substack comments sections on newsletters about freelancing

step 3: confirm they're real

check their website. if they have a website that looks like it was built in 2019 with a gmail address in the footer, they care about brand presence. good signal.

step 4: write one sentence about their specific situation

not "i noticed you work in marketing." something like "saw your post about going independent — the first 90 days of building a client list from scratch is brutal."

that's it. that's the whole first message.

numbers from doing this manually

ran this for 6 weeks targeting consultants-going-indie:

47 outreach messages sent (manual, no automation)
14 replies (29.7% reply rate)
4 paid brand audits ($99 each)
2 became repeat clients

$594 from 47 messages. not life-changing. but the rate is what matters. 8.5% conversion from message to paid is solid for cold.

the prior spray-and-pray approach: 200+ messages, 6 replies, 0 paid.

the thing everyone skips

most people spend 80% of their time writing better messages and 0% of their time building better lists.

flip it. a mediocre message to the right person at the right moment beats a perfect message to a lukewarm list every time.

if you want the templates too

built a cold outreach kit with the exact message formats i use for this — $9. but honestly the list-building is the harder and more valuable part.

the templates just make sure you don't waste the targeting work by saying something weird once you've found the right person.

running a solo consulting or productized service? curious what your current reply rates look like.

i charged $99 for a brand audit no one asked for. here's what happened.

Yanmiayn — Wed, 06 May 2026 01:03:45 +0000

i charged $99 for a brand audit no one asked for. here's what happened.

not a brag post. more of a "here's what was weird about it" post.

the setup

i'd been doing cold outreach for a few months. the usual stuff: email sequences, follow-ups, the kit i sell. worked fine. small numbers, steady.

then someone asked me if i'd look at their brand presence. not their SEO. not their traffic. just: "do we look like a real company or not."

i charged $99. took 90 minutes. sent a loom + a 1-page google doc.

they came back with 3 referrals.

so i turned it into a product.

what the audit actually is

not a 40-page consulting report. not a template filled in with their name.

it's:

10-minute loom walking through their homepage, linkedin, and one social profile
1-page doc: 3 things that are working, 3 that are hurting, 1 specific thing to fix this week
no pitch inside the doc. just the findings.

that's it. 90 minutes of work. $99.

what i learned giving away free ones

ran a 2-week test: 25 free audits, no pitch, just findings sent cold.

6 people paid $99 after receiving the free version.

the pattern was exact: every single buyer had replied to the free audit with a specific question. "you mentioned X -- can you go deeper on that?" or "what tool would fix Y?"

the people who never replied never bought. not one.

so the conversion mechanism isn't the quality of the free audit. it's whether it generates a question. if it does, they're already thinking about the next step. if it doesn't, they consumed it like content and moved on.

what i changed based on that

the free audit used to be polished. nice formatting, section headers, conclusions.

now the free one is messier on purpose. it ends mid-thought. the last section says "i'd need to dig into X to give you a real answer here."

that one change pushed replies from ~20% to ~60%.

pricing note

$99 is intentionally low. it's not consulting. it's a fast, honest read on whether your brand looks like a real company to someone encountering it cold.

a lot of founders are too close to their own stuff to see the obvious things. that's what the audit fixes.

the actual link

if you want one: brand visibility audit ($99)

if you want the cold outreach kit i mentioned (the thing that generates the leads to sell audits to): cold outreach kit ($9)

no upsell sequence after this. just the one cta that matches where you are.

i gave away free brand audits for 2 weeks. here's what actually converted.

Yanmiayn — Wed, 06 May 2026 01:00:52 +0000

i gave away free brand audits for 2 weeks. here's what actually converted.

not a growth hack. just a thing i tried because i was tired of cold emails going nowhere.

the setup

i built a brand visibility audit product. $99 on paddle. positioned at founders and small teams who want to know why they're invisible online despite doing "everything right."

problem: no one buys audits cold. you can't prove value before they pay. it's a trust cliff.

so i flipped it.

for two weeks, i ran a free tier. someone fills out a short form -- company url, 3 goals, biggest frustration. i send back a 400-word loom + a 1-page pdf with 3 actual findings. no pitch at the end, just findings.

what happened

week 1: 11 free audits sent. 0 paid conversions. felt like charity.

week 2: 14 free audits sent. 3 people replied asking about the paid version. 2 bought.

by the end of the 2 weeks: 6 paid conversions from 25 free audits. 24% conversion rate.

the ones who bought had one thing in common -- they replied to the free audit with something specific. "you said X, can you go deeper on that?" that reply was the signal.

the exact mechanism

free audit creates a real asymmetry. they now know something about their brand they didn't before. they want more.
the loom format makes it feel personal even though the audit template is 80% the same.
i never pitched in the free audit. the paid cta was in a plain ps line: "if you want the full 20-point report + a 30-min call, it's $99 here." that's it.

what didn't work

sending free audits unsolicited (people didn't open or use them)
making the free version too polished (looked like a sales doc, not genuine help)
following up more than once after the free audit (annoyed people)

the numbers

25 free audits × ~40 min each = ~17 hours of work over 2 weeks

6 paid conversions × $99 = $594

effective hourly rate: ~$35/hr

not impressive on its own. but the audit template is now a product.

i turned the whole thing into a $99 kit that teaches the process -- the form, the loom script, the pdf template, the ps cta copy. other service businesses can use it to run their own free-to-paid funnel.

if you want to try this

the brand visibility audit is live. $99 gets you the full 20-point report on your brand's online presence -- where you're leaking trust, what's fixable this week, what's structural.

brand visibility audit -- $99

no upsell after. no 6-part email sequence. just the audit.

the compounding part

every free audit i sent became a case study i can reference. every paid audit adds to a pattern library. in 6 months that pattern library is the moat, not the audit itself.

that's the play. not scale, not automation. just proof accumulation.

building yanmiayn.com -- tools for founders who do their own outreach and distribution.

the real cost of DIY outreach (and when it stops making sense)

Yanmiayn — Tue, 05 May 2026 01:58:10 +0000

this is not about cold email tools. it's about time math. most founders are spending 10x more than they need to on outreach — not in money, but in hours.

a founder i talked to last month was spending 11 hours a week on cold outreach. two replies per week. both ghosted after the first response.

when i broke down where those 11 hours actually went:

2 hours — finding and vetting prospects
5 hours — writing, rewriting, second-guessing first messages
2 hours — writing follow-ups
1 hour — tracking replies and next steps
1 hour — staring at the "compose" window trying not to sound desperate

the problem was not the 2 hours finding people. that's research, it compounds.

the problem was the 9 hours of template work. that's not a skill problem. that's a missing system problem.

the math most founders skip

scenario	hours/week	at $75/hr opportunity cost	per month
DIY from scratch every week	9	$675	$2,700
one-time template system (set up once, reuse)	1	$75	$300
difference	—	—	$2,400 saved

$2,400/month in reclaimed time. that's the math even if your hourly value is half that.

and that's not counting what actually happens when outreach works.

what "working" outreach looks like

the founders who get consistent replies are not sending smarter words. they figured out a sequence:

a trigger-based opener (something specific about the prospect, not generic praise)
a problem statement that matches the moment they're in
a bridge to what you do — one sentence, no buzzwords
a soft CTA that doesn't demand calendar access on first contact

once you have that structure locked, each message takes 4 minutes to personalize. not 40.

the 9-hour problem is almost always a "starting from blank each time" problem. systems fix blank pages.

when to build vs. buy the system

build it yourself if:

you have 6+ hours to test and iterate this week
you've already sent 100+ cold messages and know which angles get replies in your niche
you genuinely enjoy writing cold copy

buy or use a done template system if:

you need to start this week, not after a learning curve
you've tried winging it and your reply rate is under 8%
you're in a niche where the buyer has heard every pitch twice

what's actually in a good outreach kit

if you're going to invest in a system, here's what to look for (or build):

11+ tested openers across different trigger types — just-launched, fundraised, hiring, stuck, build-in-public, hated-positioning
follow-up sequences at days 3, 7, 14 that don't feel like nagging
reply handlers for "maybe later" / "how much?" / "not interested"
a prospect vetting checklist (stop messaging people who can't buy)
a tracking sheet so nothing falls through

the kit pays back its cost the first time you close a client you would have otherwise ghosted. for most people that happens in week one.

if you want the kit i actually use for my own productized services:

cold outreach kit — $29, one-time — 11 openers, follow-up sequences, reply handlers, prospect vetting checklist. no subscription, no upsell.

and if you want to fix positioning before doing outreach (outreach to a poorly-positioned brand is just faster failure):

brand visibility audit — $299 — 90-min async deep-dive, specific recommendations, not generic PDF tips.

what's your current reply rate on cold outreach? curious where people are actually landing.

3 cold email subject lines that actually get opened (and why they work)

Yanmiayn — Tue, 05 May 2026 01:45:11 +0000

3 cold email subject lines that actually get opened (and why they work)

most cold emails die at the subject line. the email itself might be good. doesn't matter. nobody opens it.

i've sent a lot of cold email. tracked every open rate, every reply, every "who is this" response. here's what i found — the subject lines that consistently outperform everything else follow three patterns.

pattern 1: the specific observation

example:
"noticed you're still using [X] for [Y]"

"your linkedin post about [topic] — quick thought"

"[company name]'s [specific page] caught my eye"

why it works:

specificity signals effort. the human brain pattern-matches "generic blast" vs "this person looked at my stuff" in about 0.3 seconds. specific = not spam. specific = maybe relevant.

the mistake most people make: they try to fake specificity. "noticed your company is growing fast" — that's not specific. that could apply to anyone.

real specificity: you actually looked at something. one thing. name it.

open rate lift vs generic subject: usually 2-3x in my tests.

pattern 2: the low-pressure question

example:
"quick question about [thing they care about]"

"is [specific thing] still a problem for you?"

"weird question — do you use [tool/process] for [outcome]?"

why it works:

a question subject line triggers the brain's open loop reflex. unanswered questions create mild cognitive discomfort. people open to close the loop.

the "weird question" prefix is counterintuitive but it works — it signals self-awareness. most cold emails take themselves too seriously. a tiny bit of self-deprecation lowers defenses.

one rule: the question has to be genuinely answerable in one sentence. if it's too broad, it reads as fake. "do you want more revenue?" = spam. "is outbound part of your Q3 plan?" = real question.

pattern 3: the named problem

example:
"the [specific pain] problem"

"when [bad outcome] keeps happening"

"[thing they dread] — a fix that worked for us"

why it works:

people don't read subject lines to find opportunities. they scan to find things relevant to problems they're already sitting with.

if your subject line names a problem they have RIGHT NOW — not a vague problem, a specific one — they open because they're hoping for a solution. you're not selling, you're answering.

the key: you have to actually know what their current problem is. this requires research. five minutes on their linkedin, their recent tweets, their job listings. job listings especially — "hiring 3 SDRs" tells you exactly what pain they're in.

what all three have in common

none of them lead with you. they all lead with them.

generic subject lines: "introducing [product]" / "partnership opportunity" / "quick chat?" — all about you.

subject lines that get opened: about their situation, their problem, their context.

the shift is mental before it's tactical. stop trying to get opens by making your pitch more interesting. start trying to understand what's already interesting to them.

the other thing nobody talks about: subject line + preview text as a unit

most email clients show 40-60 chars of preview text after the subject. that's a second subject line for free.

if your subject is "quick question about your outbound process" and your preview text is "i noticed you're hiring..." — that's a two-part hook that reinforces itself.

most people leave preview text as the first line of the email body ("hi [name], i hope this finds you..."). that's wasted space.

write the preview text on purpose. one sentence. continues the subject line's thought.

the templates (paste-ready)

version A — specific observation:

subject: [company] + [your angle] — thought

preview: noticed you [specific observation], wanted to share something

version B — named problem:

subject: the [specific pain] problem

preview: we ran into the same thing — here's what worked

version C — low-pressure question:

subject: weird question — do you still [thing]?

preview: not a pitch, genuinely curious about your process

quick note on testing

don't test 10 subject lines at once. test one variable at a time. your first test: personal vs. problem-named. send 20 emails with personal subject, 20 with problem-named, same body. look at opens after 48 hours.

most people never test because their list is too small. if you're sending to 20 people and getting 0 replies, the problem isn't the subject line — it's the targeting. fix that first.

if this was useful, i wrote a longer breakdown on cold DM angles (same principles apply to linkedin/slack) here: 5 cold DM angles that actually get replies

the full cold outreach kit (email sequences, dm templates, objection responses) is at yanmiayn.gumroad.com/l/fqqhie if you want the whole thing.

what subject line pattern works best for you? curious if the specific observation one lands differently in different industries.

5 cold DM angles that actually get replies (exact templates included)

Yanmiayn — Tue, 05 May 2026 01:30:21 +0000

5 cold DM angles that actually get replies (exact templates included)

i sent around 200 cold DMs over a 6-week stretch last year. the first 40 were embarrassing. generic intros, vague value props, "would love to connect" finishes. reply rate was around 2%.

then i stopped trying to be clever and started matching the angle to what the person was actually experiencing that day. reply rate went to 18%. here are the 5 angles that moved the needle, with the real templates.

angle 1: the just-shipped

someone just launched something. they're in a dopamine spike. they're also terrified nobody cares.

hey [name] — just saw you shipped [product]. congrats. real question: who's doing your cold outreach right now? most founders at this stage send 10 DMs, hear nothing, and assume the product is the problem. it's almost never the product. tried a bunch of different opener formats and finally found what actually converts. happy to share if useful.

why it works: you caught them at peak motivation + peak anxiety. the question "who's doing your cold outreach" forces a real answer, not a polite ignore.

tip: send within 24 hours of the launch post. after 48 hours this angle goes cold fast.

angle 2: the stuck founder

they're posting product updates but the numbers aren't moving. you can tell because the updates get quieter, shorter, more defensive.

hey [name] — been watching [product] for a few weeks. the build looks solid but traction looks stuck. in my experience the bottleneck at this stage is almost always the outreach message, not the product. happy to share what changed my reply rate if you're open to it.

why it works: you named the thing they're embarrassed about. nobody wants to admit their DMs suck. the "in my experience" framing doesn't put them on the defensive.

tip: never say "your DMs are bad." say "outreach message" and keep it about your own experience.

angle 3: build-in-public identity blur

they post every week. updates, metrics, reflections. but zero outreach. they've built an audience, not a customer base.

hey [name] — been following your build-in-public posts on [product]. you're consistent, which is rare. but i notice a pattern: lots of updates, not much outreach. building in public builds audience. cold outreach builds customers. they're different games. if you ever want to see what the outreach side looks like, dm me.

why it works: you're not pitching a product. you're naming a strategic gap they probably sense but haven't articulated. the "different games" line tends to get a response because it's true and slightly uncomfortable.

angle 4: the show IH debut window

someone just posted on indie hackers for the first time. they have 48 hours of organic attention. most waste it waiting for inbound.

hey [name] — saw your show IH post on [product]. solid debut. here's the thing: show IH gets you 48 hours of organic attention, then it's dead. founders who turn that into customers don't wait for inbound — they DM the right 50 people before the post goes cold. happy to share the targeting criteria i used if helpful.

why it works: creates real urgency without fake scarcity. the 48-hour window is genuinely real. people respond to time-sensitive truth.

angle 5: the hated-positioning reply

they posted about their positioning feeling off, or their launch flopping, or their pitch being ignored. they already admitted the problem publicly. you just have to show up with a frame.

[name] — your [post/tweet] hit different. you basically described the exact positioning problem that kills cold outreach before it starts. when your pitch is fuzzy in public, your DMs are catastrophic. spent a few weeks fixing this for my own product — the core insight is that every DM has to assume the recipient has never heard of you and doesn't owe you attention. happy to share more if useful.

why it works: you're reacting to something they said, not cold-pitching. the "i fixed this for myself" framing builds credibility without brag.

what these have in common

they're context-specific. no generic "i love your work" openers.
they create a genuine question in the reader's head ("wait, is that true?")
they offer something small first (insight, framework) before asking for anything
the ask is low-stakes: "happy to share" not "buy this now"

reply rates above 15% consistently once i stopped rotating through the same generic opener and started matching the angle to what the person was experiencing that day.

if you want all 29 angles i've tested including follow-up sequences, the full kit is on gumroad for $29: cold outreach kit. it's a doc, not a course — copy-paste ready templates with context notes for when to use each one.

the full write-up of how i built this process: yanmiayn.com/blog/2026-05-05-cold-dms-that-worked.html

5 cold DM angles that actually get replies (with the exact templates i used)

Yanmiayn — Tue, 05 May 2026 01:24:35 +0000

5 cold DM angles that actually get replies (with the exact templates i used)

i sent 200+ cold DMs last year. first 6 weeks: 2% reply rate. embarrassing. i thought the product was the problem.

it wasn't. it was the angle. every message i sent was basically "hey, i built a thing, want it?" — dressed up in slightly different words. founders smell that from a mile away. what changed my numbers wasn't better copy per se, it was understanding what situation the person is in and matching the opener to that situation.

went from 2% to 18% reply rate. not by being clever. by being specific.

here are the 5 angles i actually use, with the real templates:

angle 1: the just-shipped

who it's for: founders who launched in the last 72 hours. they're on launch high, watching their metrics, hyper-receptive.

the template:

hey [name] — just saw you shipped [product]. congrats. real question: who's doing your cold outreach right now? most founders at this stage send 10 DMs, hear nothing, and assume the product is the problem. it's almost never the product. the outreach angle is usually what's broken.

why it works: you're meeting them at the exact moment they're thinking about growth. the "real question" framing signals you're not pitching yet. you're diagnosing.

what not to do: don't say "love what you're building." they've heard it 40 times today.

angle 2: the stuck founder

who it's for: someone who shipped weeks ago, has a product page, but the twitter/X updates have gone quiet. low-engagement posts. the vibe of "hmm, this isn't working."

the template:

hey [name] — [product] looks solid. but i notice the traction isn't matching the build quality. most founders at this stage are sending weak cold messages — the kind that feel like form letters. i went through the same thing and spent 3 weeks fixing it. happy to share what worked if that's where you're stuck.

why it works: you're naming the situation without them having to admit it. "traction isn't matching build quality" is flattering but honest. it opens a conversation.

what not to do: don't open with "i noticed your engagement is low." that stings. lead with the product quality observation first.

angle 3: build-in-public identity blur

who it's for: founders who post daily about their build, have 500-2k followers, but have publicly mentioned "no paying users" or "struggling to convert."

the template:

hey [name] — been following your build-in-public posts on [product]. you're consistent, which is rare. but i notice a pattern: lots of updates, not much outreach. building in public builds audience. cold outreach builds customers. they're different games.

why it works: you're affirming their consistency (genuine) then pointing at the actual gap. it's not a dunk. it's a reframe. and founders who build in public are already comfortable with feedback loops.

what not to do: don't pitch on the first message. this angle is a conversation starter. let them respond.

angle 4: the show IH / show HN debut

who it's for: someone who just posted "Show IH" or "Show HN." they have 48 hours of organic attention.

the template:

hey [name] — saw your Show IH post on [product]. solid debut. here's the thing: Show IH gets you 48 hours of organic attention, then it's dead. the founders who turn that into actual customers don't wait for inbound — they DM the right 50 people before the post goes cold. just thought i'd flag it.

why it works: urgency is real, not manufactured. the 48-hour window actually exists. you're giving them useful information whether or not they ever talk to you again.

angle 5: the bad-positioning recovery

who it's for: founders whose homepage copy is generic ("the all-in-one solution for..."). their cold messages probably mirror the same vague positioning.

the template:

hey [name] — checked out [product]. the build is real but the hero copy is doing you dirty in cold outreach. founders who copy-paste landing page positioning into DMs see reply rates collapse. different context, different psychology. happy to show what i mean with one specific example from your current copy.

why it works: you're offering something specific and demonstrable. "one specific example from your current copy" is a low-commitment offer that's hard to say no to.

what to do after they reply

the reply is not the win. the reply is the start. most founders celebrate the response and then send a clunky 200-word pitch. don't.

mirror their energy. if they write 2 sentences, write 2 sentences. ask one question. stay in the conversation.

the follow-up timing matters too: day 1 is send, day 4 is follow-up #1 ("did this land in spam or just not relevant?" — 7 words, no pressure), day 8 is the last touch if nothing.

if you want all 29 angles + the follow-up sequences

i packaged the full system — 29 DM scripts organized by founder situation, plus the follow-up cadence for each angle. $29 one-time, instant download.

wrote up the full process here: yanmiayn.com/blog/2026-05-05-cold-dms-that-worked.html

or get the kit directly on gumroad: yanmiayn.gumroad.com/l/fqqhie — $29, no subscription.

shipped this after 3 months of iterating on real outreach. nothing theoretical. these are the exact words i used.

I redesigned my SaaS hub: every product card is a pure-CSS animation of what it does

Yanmiayn — Sun, 03 May 2026 23:34:20 +0000

I just redesigned yanmiayn.com — my one-person SaaS hub — and the part I'm proudest of is that every product card on the homepage is a pure-CSS animation of what the product actually does.

Hover the cards on the page to see them in motion. No SVGs, no JS for the animations themselves — only transform, keyframes, and cubic-bezier.

The four cards

1. Consensus Cold Outreach Worker — $29 ($0 → $10K MRR public log)
6 small dots arranged in a circle (each = one LLM). They collapse toward the center. One dot — the "winner" — rises in the middle, full opacity, scaled up. Loop.

2. Infpilntr — Parallel RFQ
A single source dot on the left. Three legs fan out (rotated -26°, 0°, +26°) with scaleX from 0 → 1, staggered by 120ms. Each lands on a far-right end dot.

3. DocChase — Doc intake
Three "papers" stack at the top. Each translateY(70px) rotate(2deg) scale(0.6) with cubic-bezier(0.34, 1.56, 0.64, 1). Staggered 1.2s apart. They literally slide into a folder shape (CSS ::before for the tab).

4. OpenClaw Consensus API — 9 LLMs in one endpoint
Nine dots sit on a circle (rotate(40°i) translateY(-56px)). The whole ring breathes — radius 64 → 42 → 64 — with a center "core" pulsing in inverse rhythm.

Stack

Tailwind CDN
Inter + JetBrains Mono
Vanilla CSS keyframes (no library)
Express + Resend on Hetzner for the email subscribe form (working — /api/subscribe posts to a JSON line file + sends a welcome via Resend)

The "honest" part

Grayscale only. No orange CTAs, no rose badges, no violet pills. Zinc tokens 50–900.
Two subscribe forms (hero + footer) — ranked by where attention actually lands.
Email actually sends. From hello@yanmiayn.com (DKIM + SPF verified on Resend).

If you build similar one-person stacks, I'd love brutal feedback on the conversion path: hover → product page → Gumroad checkout. Or if any of the keyframes are uglier than they look on hover, please tell me.

Live: yanmiayn.com

Source for the four animations is inlined in the homepage <style> block — view-source:https://www.yanmiayn.com/ and search for a-vote, a-split, a-doc, a-breathe.

— solo founder, public log in progress

I Built a 24/7 AI Agent System on a $6/Month VPS — Here's the Stack

Yanmiayn — Wed, 29 Apr 2026 21:25:02 +0000

The Setup

Running a fully autonomous AI agent on Hetzner VPS (€3.9/mo).

Stack:

OpenClaw agent framework (open source, MIT)
DeepSeek V4 Pro via NVIDIA NIM API (1.6T MoE, 1M context)
Playwright + Python for browser automation
Docker container for isolation

What it does autonomously:

Posts content to Twitter/X every 5 minutes
Publishes Dev.to articles every 30 minutes
Manages Gumroad digital product store
Sends targeted promotions

Why DeepSeek V4 Pro?

Released April 24, 2026:

1M token context (real, tested)
$1.74/1M input vs Claude $3
Think/Non-Think dual modes
MIT license

5x cheaper than Claude Sonnet 4.6 for typical agent workloads.

The Result

89 AI productivity guides packaged and selling on autopilot.

Full setup playbooks: https://yanmiayn.gumroad.com

What automation are you building?