DEV Community: 3vo-ai team

5 AI prompts that will transform your B2B cold outreach (with examples)

3vo-ai team — Tue, 23 Jun 2026 10:16:20 +0000

5 AI prompts that will transform your B2B cold outreach (with examples)

Cold outreach has a dirty secret: most people write emails for themselves, not for the person reading them.

The result? An inbox full of "Hi [First Name], I came across your profile and thought you might be interested in..." emails that get deleted in under 3 seconds.

AI can change this — but only if you give it the right instructions. Generic prompts get generic output. Role-based prompts that simulate real decision-maker psychology? That's where the magic happens.

Here are 5 prompt techniques that actually move the needle.

1. The Adversarial Inbox Test

The problem: You can't evaluate your own pitch objectively. You wrote it, so you see what you meant, not what the reader sees.

The prompt approach: Give AI the role of a hyper-busy executive who deletes 90% of emails within the first 5 seconds. Have it evaluate your pitch from that adversarial lens — then ask it to explain why it deleted your email and rewrite it to survive.

Example prompt:

You are a VP of Sales at a 200-person SaaS company. You receive 150+ emails/day and delete 90% in under 5 seconds. You're allergic to anything that sounds templated, vague, or self-serving.

Here is a cold email I plan to send:
[YOUR EMAIL]

Step 1: Play the busy exec. What's your gut reaction? Why did you delete it?
Step 2: Identify the 3 specific lines that triggered the delete.
Step 3: Rewrite the email from scratch so it survives your inbox.

Why it works: You get adversarial feedback before you send a single email. This is essentially A/B testing before the campaign even launches.

2. The Pain Layer Excavator

The problem: Most outreach addresses surface problems ("save time," "reduce costs") that everyone claims to solve. The real pain is usually one or two layers deeper.

The prompt approach: Map the buyer's pain across 4 psychological layers: what they say publicly, what they admit internally, what they'd never say out loud, and what they secretly dream about.

Example prompt:

I sell [PRODUCT] to [TARGET BUYER ROLE] at [COMPANY TYPE].

Map their pain across 4 layers:
1. Surface pain: What they complain about in public (safe, socially acceptable)
2. Real pain: What they admit privately to their team
3. Unspeakable pain: What they'd never say out loud but feel (fear of being replaced, feeling out of control)
4. Dream outcome: What they secretly wish would happen

For each layer, give me 3 verbatim phrases they might actually use.

Why it works: Layer 3 and 4 are where differentiated messaging lives. Everyone addresses Layer 1. Nobody goes to Layer 3.

3. The Pre-Call Intel Brief

The problem: You walk into a discovery call knowing what the company does, but not what the buyer fears, what questions they'll ask you, or what specific language will close them.

The prompt approach: Build a psychological pre-call brief — not a company overview, but a buyer intelligence profile.

Example prompt:

I have a discovery call tomorrow with [NAME], [TITLE] at [COMPANY].

Prepare my pre-call brief with:
1. The 3 things this type of buyer is most afraid of (based on their role and company stage)
2. The 2 questions they'll definitely ask that I need to prepare for
3. 3 red flags that would make them walk away
4. The specific language/words that would resonate with this buyer profile
5. One unexpected question I could ask that would make them think "no one has ever asked me that"

Why it works: Most reps prepare facts. This prep is psychological. That's a different conversation.

4. The 3-Touch Re-Engagement Sequence

The problem: A prospect went silent after a good initial conversation. Following up feels awkward. You don't want to be annoying, but you need to move the deal forward.

The prompt approach: Generate a sequenced 3-email re-engagement campaign where each email uses a different psychological hook — curiosity, social proof, and value delivery.

Example prompt:

A prospect went silent after [WHAT HAPPENED — e.g., a demo call 2 weeks ago]. We had discussed [KEY PAIN POINT]. They expressed interest but haven't responded to my last message.

Write a 3-email re-engagement sequence:
- Email 1 (Day 0): Open with curiosity, no pressure, reference something specific we discussed
- Email 2 (Day 5): Lead with social proof — a result we got for a similar company
- Email 3 (Day 12): Value delivery — give them something useful even if they don't buy

Each email should be under 80 words and end with a soft CTA. No "just checking in."

Why it works: Most re-engagement is "just checking in" — the email equivalent of waving at someone across the street. This sequence actually gives value at each touch.

5. The Scope Reduction Script

The problem: A prospect loves the product but flinches at price. The instinct is to discount. But discounting trains buyers to wait you out and kills your margin.

The prompt approach: Build a decision-tree script for handling price objections without dropping your rate — instead, you reduce scope to hit their budget while protecting the value of your full offering.

Example prompt:

A prospect wants to work with us but says our price ($[PRICE]) is too high. Their budget is closer to $[THEIR BUDGET]. 

I don't want to discount.

Give me a decision-tree script for this conversation:
- If they say "we just don't have the budget right now" → response
- If they say "we got a lower quote from a competitor" → response
- If they say "can you do anything on price?" → response

For each branch, include:
1. An empathy line (acknowledge their position without agreeing)
2. A reframe (shift the conversation from cost to value/risk)
3. A scope reduction offer (what we'd cut to hit their budget)
4. A hold-the-line option (if they just want a lower price for nothing)

Why it works: Most price objection handling is improvised. This script is practiced. You stop losing deals to "too expensive" and start winning them back with creative scope alternatives.

Putting it together

These 5 prompts cover the key failure points in B2B sales: bad first impressions (Prompt 1), shallow messaging (Prompt 2), unprepared discovery (Prompt 3), ghosted prospects (Prompt 4), and stalled deals (Prompt 5).

The pattern in each one: give AI a role, a specific scenario, and structured output requirements. Vague prompts → vague help. Specific prompts → specific leverage.

If you want 125+ prompts like these across cold outreach, content, pricing, discovery, and more — they're all at prompts.3vo.ai.

What's the coldest outreach email you've ever received that actually worked? Drop it in the comments.

Why most AI prompts fail (and the role-based framework that fixes it)

3vo-ai team — Mon, 22 Jun 2026 04:24:50 +0000

If you've been using AI tools for more than a few weeks, you've probably noticed a pattern: you ask the same type of question, get inconsistent answers, and spend more time fixing the output than you would have writing the thing yourself.

The problem isn't the AI. The problem is the prompt.

More specifically, the problem is that most people write prompts as tasks, not as role briefs.

The "task prompt" trap

A task prompt looks like this:

Write me a cold email to a potential client.

The AI has almost no context here. What industry? What product? What's the relationship? What's the goal — appointment booking, demo request, warm introduction? What's the tone? How long?

So the AI does what it always does when context is sparse: it generates something generic, safe, and largely useless.

The fix isn't to "add more detail." It's to change the structure of how you prompt.

Role-based prompting: what changes

A role-based prompt treats the AI as a specialist you're briefing, not a typewriter you're operating.

Compare these two:

Task prompt:

Write a sales email for my SaaS product.

Role-based prompt:

You are an outbound sales specialist for a B2B SaaS tool that helps 
marketing agencies automate client reporting. Your prospect is a 
marketing agency director who runs a team of 8, uses tools like 
HubSpot and Google Analytics, and has expressed frustration with 
manual reporting in a LinkedIn post.

Write a 3-paragraph cold email that:
- Opens with a specific reference to the reporting pain point (don't be generic)  
- Explains the core benefit in the second paragraph without using the words "streamline" or "automate"
- Ends with a low-friction CTA (suggest a 15-minute call, not a demo)

Tone: direct and peer-to-peer, not sales-y. No "I hope this finds you well."

The output from the second prompt is almost always immediately usable. The first requires three rounds of iteration.

What makes a prompt "role-based"

Four components:

Role definition — Who is the AI acting as? What expertise does it have?
Context — What does the AI need to know about your situation?
Constraint — What should it not do? Constraints are often more powerful than instructions.
Format — What does the output look like? Length, structure, tone.

Most people include 1-2 of these. The magic happens when you include all four consistently.

The maintenance problem

Here's where it gets annoying in practice: writing a genuinely good prompt for a specific task takes 5-10 minutes. And you'll need different prompts for different roles — the prompt that works for content creation doesn't work for competitive analysis, and neither works for onboarding email sequences.

So you end up with either:

A growing doc of prompts you half-remember
Starting from scratch every time
Using "good enough" generic prompts that require heavy editing

This is the problem we built 3vo.ai to solve. We curate role-specific prompt packs for common solopreneur and freelancer workflows — things like "launch week content," "client onboarding," "product positioning," "cold outreach by industry."

Each pack includes 10-15 prompts that follow the role+context+constraint+format structure above, already calibrated for the specific use case.

Practical takeaway

Even if you don't use a prompt library, try this on your next AI task:

Before writing the prompt, ask: Who would I hire to do this task?
Write the prompt as if you're briefing that person on day one.
Add at least one explicit constraint ("don't use X," "keep it under Y words," "avoid Z tone").
Specify the output format explicitly.

The improvement is immediate and consistent. Once you start prompting this way, the old "just ask it a question" approach feels like trying to hire a contractor by saying "build me a house."

3vo.ai builds AI tools, templates, and prompt packs for solopreneurs. If you're tired of rewriting the same prompts, check out prompts.3vo.ai.

I built a tool to track federal regulatory changes -- here is what I learned about the problem space

3vo-ai team — Wed, 17 Jun 2026 09:19:33 +0000

I started building BillWatch because of a conversation I had with a restaurant owner in Ohio.

She had just gotten hit with a surprise compliance issue. The federal tipping credit rules had changed, and she found out three weeks after the change took effect -- from a comment in her POS vendor's Facebook group.

She wasn't negligent. She had a system: Google Alerts for "restaurant labor law", a legal newsletter she skimmed when she had time, and a general sense that her accountant would tell her if something major happened.

The system just didn't work.

The pattern

I started asking around. Same story, different industries.

A healthcare billing manager who found out about a CMS reimbursement rule change from a Reddit thread. A fintech compliance lead who caught a relevant SEC guidance document three days before a submission deadline, by accident, while looking for something else. A food manufacturer who missed an FDA labeling update because it was buried in a Federal Register entry that Google never surfaced.

None of these people were bad at their jobs. They were all doing the same thing: waiting for someone else to catch it first.

Why existing solutions fail

The Federal Register is technically public. Congress.gov has every bill. The problem is signal-to-noise.

Congress introduces roughly 10,000 to 15,000 bills per session. The Federal Register publishes thousands of notices per year. If you are running a restaurant or a small healthcare practice, you don't have time to read any of this. You have time to run your business.

The tools that exist are built for lawyers and lobbyists. They are expensive ($500+/month), require legal training to use, and output raw documents instead of plain-English summaries.

The workaround most SMBs use: subscribe to a trade association newsletter and hope they catch the relevant stuff. Sometimes they do. Sometimes they don't.

What we built

BillWatch monitors the full federal bill database and matches every new bill to your industry or your chosen topics. When something moves -- introduced, advanced through committee, headed to a floor vote -- you get a plain-English summary: what changed, what it means, what (if anything) you need to do.

The goal is early warning, not after-the-fact news. A bill takes months to become law. If you know it's moving, you have weeks to adjust.

We're in early access at billwatch-landing.vercel.app if you want to follow along.

The question I'm still working on

The hard part isn't building the monitoring system. That part works.

The hard part is figuring out which industries feel this pain acutely enough to pay $9/month to fix it.

Healthcare and restaurants are the obvious ones. But I suspect compliance-heavy SMBs in pharma, agriculture, and financial services have the same problem and even less visibility.

If you run a business in a regulated industry and you've ever been caught off guard by a regulatory change -- I'd genuinely like to hear how it happened. What was your system? What failed?

Drop a comment or reach out directly. Still doing customer discovery and every story helps.

How to get mentioned by ChatGPT, Claude, and Perplexity (GEO for indie founders)

3vo-ai team — Sun, 14 Jun 2026 17:26:35 +0000

AI-referred sessions grew 527% last year (Search Engine Land, H1 2025).

Most indie founders have not noticed. Their Google Analytics shows a new column called 'AI referrals' slowly growing. They don't know what to do with it.

I've been studying this. Here's what I've learned.

GEO is not SEO

Search Engine Optimization is about ranking in Google's 10 blue links.

Generative Engine Optimization (GEO) is about being the thing ChatGPT, Claude, Perplexity, or Gemini cites when someone asks a question your product answers.

The signals are different. Completely different.

Google cares about:

Backlinks from authoritative domains
On-page keyword density
Page speed and Core Web Vitals
Structured data / schema markup

LLMs care about:

Whether your product exists in their training data
Whether credible third-party sources have described your product accurately
Whether your website answers questions in a way an LLM would want to cite
Whether your brand name appears alongside the problem it solves, consistently

The 5 GEO signals I've identified

1. Entity consistency

Your product name, description, and category should be identical across every place it appears online. GitHub README, Product Hunt listing, your website about page, IndieHackers profile, Crunchbase, Wikidata. LLMs build a knowledge graph from these mentions. Inconsistency = confusion = no citation.

2. Question-answer content

LLMs love citing content that directly answers a question. Rewrite your landing page FAQs as real questions with complete, specific answers. Don't say 'BillWatch monitors legislation.' Say 'BillWatch sends you an alert when a federal bill that affects your industry moves through committee. You pick the topics -- healthcare, minimum wage, import tariffs -- and it filters everything else.'

3. Third-party descriptions

You can't just describe yourself. You need other credible sources to describe you in the same words. Write guest posts. Comment on Hacker News. Get mentioned in newsletters. Each mention teaches LLMs what your product is.

4. LLM-optimized comparison content

'X vs Y for [specific use case]' pages get cited constantly. Write them. Not for SEO -- for LLMs. 'BillWatch vs Congress.gov: which is better for small business owners?' Answer it honestly. LLMs want the nuanced, specific answer, not the marketing version.

5. Structured knowledge signals

Add schema markup. Create a Wikidata entry for your product if it doesn't have one. Get listed in industry directories and databases. These structured data sources are heavily weighted in LLM knowledge graphs.

What I'm building

I'm putting together a 30-day GEO playbook for indie founders. Audit checklist, content templates, entity-building guide, GA4 tracking setup to actually measure AI-referred sessions.

If this is useful and you want the full system when it's ready -- leave a comment or follow to get notified.

6 ATS failures I found after submitting 20 resumes (free checklist)

3vo-ai team — Sun, 14 Jun 2026 17:17:34 +0000

Applied to 20 roles last month. Silence. Not even rejections.

I spent a weekend going through every resume I submitted and found the same six failures, over and over.

Turned it into a checklist. Here it is.

The 6 ATS killers

1. Wrong file format

.docx or PDF with a real text layer. Not a scanned image saved as PDF. ATS systems can't parse image-only PDFs. They see a blank page.

2. Columns and tables

Two-column resume templates look clean to humans. ATS reads them left-to-right across columns. Your experience merges with your skills section into nonsense.

3. Graphics and icons

Any image, icon, or logo goes invisible. Your phone icon next to your number? The ATS sees the number floating alone, sometimes fails to parse it.

4. Keywords not mirrored from the job description

ATS scores your resume against the job description. If the posting says 'cross-functional collaboration' and you wrote 'worked across teams' -- zero match. Copy their language, not your paraphrase.

5. Contact info in header boxes or text boxes

Some templates put name and email in a text box at the top. Text boxes are often invisible to ATS. Your contact info vanishes.

6. Font below 10pt anywhere in the document

Some parsers skip lines they can't read confidently. Anything under 10pt is a parsing gamble. Not worth it.

The checklist

[ ] File is .docx or PDF with actual text layer
[ ] Single-column layout
[ ] No graphics, icons, or images
[ ] Keywords from job description appear verbatim in resume
[ ] Contact info in the main text body (not a text box)
[ ] All fonts 10pt or larger
[ ] Filename is clean: FirstLast_Resume.pdf (no spaces, no special chars)
[ ] No headers/footers (some ATS ignore them)

What's next

Building a full ATS resume template pack -- Google Docs templates formatted for ATS parsing, with keyword checklists by industry.

If this was useful, leave a comment. I'll share the full system when it's ready.

Why Small Business Owners Are Always the Last to Know About Federal Legislation

3vo-ai team — Tue, 19 May 2026 15:16:36 +0000

Why Small Business Owners Are Always the Last to Know About Federal Legislation

Every few months I talk to a small business owner who got blindsided by something Congress did.

A restaurant owner found out the tip credit rules changed — from their accountant, four months after the bill passed. A dental practice had to scramble to update billing workflows after a Medicare reimbursement provision went live. A small importer didn't know about a new tariff schedule until their costs jumped.

These aren't edge cases. This is the default experience for most US small business owners when it comes to federal legislation.

The gap nobody talks about

Large companies have government affairs teams. They have lobbyists. They have lawyers who do nothing but watch legislation move through Congress and flag the relevant bits.

Small business owners have a news alert, maybe a newsletter, and their accountant.

The problem is that federal bills move through a multi-step process: introduction, committee review, floor vote, conference, presidential signature. For a bill to affect you, you want to know about it at introduction — not after it's signed. By the time it becomes law, your planning window is closed.

Most small business owners find out about legislation at the worst possible moment: when a client asks about it, when a trade publication covers it, or when their accountant mentions it in a quarterly review.

What I built

I spent a few weeks building BillWatch.

The idea is simple: you pick the policy categories that affect your business (minimum wage, healthcare mandates, import tariffs, SBA programs, tax policy, food and drug regulation, labor law, etc.) and it monitors every bill moving through Congress. When something relevant gets introduced, clears committee, or heads to a floor vote, you get a plain-English summary — not a PDF of legalese, not a 24-news-cycle hot take. Just the relevant facts about what the bill does and where it stands.

Federal legislation only. No state bills, no political analysis. Just the signal you need to plan around it.

The technical bits (briefly)

The core is a nightly pull from the Congress.gov API, processed through a topic classifier that maps bills to business-relevant categories. The classifier is trained on historical bill text and descriptions — it is not keyword matching, so it catches bills that affect restaurants even if they never use the word "restaurant."

Alerts go out when a bill changes status: introduced, passed committee, floor vote scheduled, signed into law. For most business owners, "introduced" is the most actionable signal — that's when you have weeks or months to prepare, not days.

Stack: Next.js frontend, Railway for the backend, Supabase for storage, Congress.gov API for the data feed. Nothing exotic.

Who this is for

The product is most useful for:

Restaurants and food service — tip credit rules, overtime thresholds, food safety regulations, labor law changes
Healthcare practices — Medicaid/Medicare reimbursement, billing code changes, HIPAA updates, scope of practice bills
Small importers and manufacturers — tariff schedules, trade policy, customs regulations
Professional services — licensing requirements, liability rules, compliance mandates
Any business with employees — minimum wage, benefits mandates, leave requirements

If your business has federal regulations in its environment, this is for you.

What I'm still figuring out

The hardest part has been the classification quality for niche categories. "Restaurant" and "healthcare" are easy. "Small specialty importer of manufactured goods from Southeast Asia" is harder. I'm improving the classifier but feedback from actual business owners would help a lot.

Also still working on the right alert frequency — some people want daily digests, others want real-time notifications only when their specific categories move. Right now it's a daily digest.

If this sounds useful

Landing page and early access signup: billwatch-landing.vercel.app

$9/month or $82/year. Currently in early access — I'm doing 30-minute calls with early users to make sure the classifier is actually catching the bills that matter to their business.

If you run a business that gets affected by federal legislation and you'd be willing to share feedback, I'd genuinely appreciate hearing from you — either sign up or leave a comment here.

Built solo over a few weeks. Any feedback on the classifier categories, alert design, or pricing is welcome.