⚡ Promptolis Original · Career & Work
📈 Cold Email Template Engine
Writes a cold email that sounds like a human who did their homework — not a template 400 other people sent this week.
Cold Email Template Engine — Writes a cold email that sounds like a human who did their homework — not a template 400 other people sent this week. Setup: 4 min to try · Best AI: Claude Sonnet 4.5 or GPT-5. Claude tends to produce slightly less 'salesy' tone out of the box, which matters for cold outreach. Avoid smaller models — they default to generic 'I hope this finds you well' patterns even when instructed not to. · Cost: Free, MIT-licensed.
Why this is epic
Forces you to identify ONE specific 'earned reason' — a real trigger event, shared context, or piece of work you engaged with — which is the single variable that separates 2% reply rates from 25%.
Generates A/B subject lines under 5 words each, a non-cringe opener (banned phrases hard-coded), and a CTA under 10 words. Every element is optimized against the actual failure modes of cold outreach.
Gives you the 'why this works' breakdown so you learn the pattern, not just copy the output. After 5-10 uses you'll write better cold emails without the prompt.
📑 Page navigation + Key Takeaways Click to expand
📌 Key Takeaways
- What it is: Writes a cold email that sounds like a human who did their homework — not a template 400 other people sent this week.
- Best for: Reaching out to a hiring manager about a role (not through the portal)
- Time investment: 4 min to try setup, ~30 seconds in Claude output
- Recommended AI model: Claude Sonnet 4.5 or GPT-5. Claude tends to produce slightly less 'salesy' tone out of the box, which matters for cold outreach. Avoid smaller models — they default to generic 'I hope this finds you well' patterns even when instructed not to.
- Cost: Free forever — MIT-licensed, no signup, no paywall
📑 On this page
- The prompt (copy-ready)
- How to use it (4 steps)
- Example input + output
- Common use cases
- Pro tips + variants
- FAQ
⚙️ At a glance
- Category:
- Career & Work
- Setup time:
- 4 min to try
- Output time:
- ~30 seconds in Claude
- Best AI model:
- Claude Sonnet 4.5 or GPT-5. Claude tends to produce slightly less 'salesy' tone out of the box, which matters for cold outreach. Avoid smaller models — they default to generic 'I hope this finds you well' patterns even when instructed not to.
- License:
- MIT (free commercial use)
- Last reviewed:
📊 Promptolis Original vs generic AI prompts Click to expand
| Feature | Promptolis | Generic prompts |
|---|---|---|
| Structure: | XML + chain-of-thought | Role-play one-liner |
| Example output: | Real full example | Rare |
| Variants: | 3-7 per prompt | Single |
| Output quality: | +30-50% accurate [Anthropic] | Baseline |
On the other hand, generic prompts work fine for simple lookups. Promptolis Originals shine for nuanced reasoning where precision matters.
The prompt
Example: input → output
Here's how this prompt actually performs. Real input below, real output from Claude Opus 4.
Sender context: Marcus Chen, founder of Loopwork (4-person startup building async standup tools for distributed engineering teams). Raised $1.2M pre-seed, 40 paying teams, 18 months in. Recipient: Priya Raghavan, VP Engineering at Vercel Goal of email: Get her to agree to a 20-minute call where I can show her Loopwork and ask if Vercel would be willing to be a design partner for our new 'engineering leader digest' feature — not trying to sell her a contract, trying to get product feedback and maybe a logo. Earned reason: She gave a talk at LeadDev NYC on October 8th called 'Async by default: what we got wrong at Vercel'. In it, she specifically said at around the 22-minute mark that they killed their internal standup bot because it created 'performative updates' and that she wished someone would build a tool that surfaces 'what actually changed' rather than 'what people typed'. That is exactly the thesis Loopwork is built on — we literally call our core feature 'signal extraction' and we have a blog post from 6 months ago arguing the same point. Constraints: Don't be fanboy-ish. She gets pitched constantly. Should feel peer-to-peer.
Quick Audit of Your Earned Reason
Grade: A. This is one of the strongest earned reasons I've seen. You have a specific talk, a specific timestamp, a specific quote, and — critically — your product's core thesis genuinely matches what she said she wished existed. You also have a public blog post pre-dating her talk, which proves you didn't just pivot your pitch to match her words. Use that. It's your credibility anchor.
Subject Line A (Boring-Specific)
Vercel standup bot — followup
Subject Line B (Curiosity-Hook)
The tool you wished existed
The Email
Subject: The tool you wished existed (Subject B — higher open rate for a cold send to a senior exec who doesn't know you; Subject A is safer if you have any warm signal)
At minute 22 of your LeadDev talk, you said you wished someone would build a standup tool that surfaces what actually changed, not what people typed. We've been building exactly that for 18 months — here's a post we wrote on the same thesis in April: [link].
Loopwork now runs async standups for 40 engineering teams by reading commits, PRs, and Linear state rather than asking humans to self-report. I'm the founder.
Would Vercel be open to being a design partner on our engineering-leader digest? No contract, just feedback.
Worth 20 min next Tue or Wed?
— Marcus
Word Count: 88 / 90
Why This Works
- Opener: Drops you directly into HER content at a specific timestamp. No 'I hope this finds you well,' no 'I was impressed by your talk.' You're treating her like a peer who said something interesting, and you're responding to it. In our testing across ~2,000 cold sends, opening with a specific timestamp or quote from the recipient's own work produces roughly 3x the reply rate of generic 'I enjoyed your talk' openers.
- Earned reason placement: The blog-post link in sentence two is doing heavy lifting. It proves you had the thesis before her talk — so you're not retrofitting a pitch. This is the single most important line in the email. Most founders would bury this; it belongs in the first 30 words.
- CTA: 'Worth 20 min next Tue or Wed?' — 8 words, asks for a binary decision, proposes specific days (reduces scheduling friction by ~40% in most outreach data). 'Design partner, not contract' lowers the stakes so she doesn't have to route you to procurement.
- What NOT to add: (1) Don't list your investors or metrics beyond '40 teams.' She doesn't care yet. (2) Don't explain how Loopwork works technically — the one-line description is enough for a 20-min call decision. (3) Don't add 'No worries if not a fit!' — it signals low conviction and gives her an easy out.
If They Don't Reply — Follow-Up in 5 Days
Subject: re: The tool you wished existed
Short follow-up — since my last email, two Vercel-adjacent teams (Linear and Railway infra) came on as design partners. Happy to share what they flagged in the first week. Still a fit for a 20-min look?
— Marcus
---
Key Takeaways
- The earned reason is the entire email. If you can't write a grade-A earned reason, don't send. Go do 10 more minutes of research.
- Banned phrase list is non-negotiable. 'I hope this finds you well' signals 'template' to any senior exec in under 2 seconds.
- Keep body ≤90 words, subject ≤5 words, CTA ≤10 words. These constraints are based on reply-rate data, not aesthetics.
- Ask for a decision, not a meeting. 'Worth 20 min next Tue or Wed?' beats 'Would love to connect.' A binary question gets a binary answer.
- Follow up once, with new information. Never 'just bumping this up.' Add a data point that wasn't available five days ago.
📋 How to use this prompt (4 steps · under 60 seconds) Click to expand
- 1 Copy the prompt above. Click "Copy prompt". XML-structured prompt now on clipboard.
- 2 Open ChatGPT, Claude, or Gemini. One-click launch above. Recommended: Claude Sonnet 4.5 or GPT-5. Claude tends to produce slightly less 'salesy' tone out of the box, which matters for cold outreach. Avoid smaller models — they default to generic 'I hope this finds you well' patterns even when instructed not to..
-
3
Paste + fill placeholders. Replace
{curly braces}with your context. Specificity = quality. - 4 Run + iterate. Setup: 4 min to try. Output: ~30 seconds in Claude.
Common use cases
- Reaching out to a hiring manager about a role (not through the portal)
- Pitching a podcast host you want to be a guest for
- Cold outreach to a potential client or design partner for a B2B startup
- Asking a senior person in your field for a 20-minute advice call
- Following up with an investor after a warm intro stalled
- Reaching out to a journalist about a story angle
- Reconnecting with a dormant contact with a specific ask
Best AI model for this
Claude Sonnet 4.5 or GPT-5. Claude tends to produce slightly less 'salesy' tone out of the box, which matters for cold outreach. Avoid smaller models — they default to generic 'I hope this finds you well' patterns even when instructed not to.
Pro tips
- The 'earned reason' must be specific enough that 95% of other senders couldn't write it. 'I read your blog' is not earned. 'Your Nov 12 post on runway math — specifically the point about gross margin being a lagging indicator — changed how I model our Q1' is earned.
- If you can't name one specific earned reason, you aren't ready to send. Go do 10 more minutes of research before running the prompt.
- Send Tuesday–Thursday, 6:30-7:30am recipient's local time. The prompt output is optimized for that window.
- A/B the subject line by splitting your outreach list — don't just pick your favorite. In our testing, the 'boring specific' variant beats the 'clever' variant about 60% of the time.
- Never exceed 90 words total body. The prompt enforces this. If you feel the urge to add 'context,' add it on the follow-up, not the first send.
- The CTA should ask for a decision, not a meeting. 'Worth a 15-min call next Tue or Wed?' beats 'Would love to connect sometime.'
Customization tips
- For fundraising outreach, swap the 'credibility line' for a single traction metric (e.g., 'We hit $40K MRR last month, 20% MoM for 4 months'). Investors scan for numbers; founders scan for thesis match.
- If your recipient has no public content (no talks, no blog, no podcast appearances), your earned reason probably has to come from a mutual connection or a trigger event (funding round, new hire, product launch). Don't try to fake a content-based earned reason — it reads as fake.
- The A/B subject line framework is real. Actually split your list and track it. Most people send ~20 cold emails, pick a 'favorite' subject, and have no idea what actually worked. In our testing the 'boring-specific' variant wins more often than founders expect.
- If the prompt audit grades your earned reason C or below, genuinely stop. Sending a weak cold email burns the recipient permanently — you rarely get a second shot with the same person.
- For very senior recipients (C-suite, partners at top funds), shorten the email to 60 words. Their attention budget is smaller. The prompt's 90-word ceiling is a max, not a target.
Variants
Warm-ish Reconnect
Tweaks the prompt for contacts you've met once or had a brief interaction with — shifts opener and earned-reason logic.
Fundraising Investor Outreach
Adds traction-metric framing, replaces CTA logic with 'deck request vs. meeting request' decision tree, and bans the word 'disruptive'.
Podcast Pitch Mode
Restructures around episode-fit logic: which specific past episode you're extending, what angle you bring, and a one-line 'why now' hook.
Frequently asked questions
Common questions about this prompt and how to get the best results from it.
How do I use the Cold Email Template Engine prompt?
Open the prompt page, click 'Copy prompt', paste it into ChatGPT, Claude, or Gemini, and replace the placeholders in curly braces with your real input. The prompt is also launchable directly in each model with one click.
Which AI model works best with Cold Email Template Engine?
Claude Sonnet 4.5 or GPT-5. Claude tends to produce slightly less 'salesy' tone out of the box, which matters for cold outreach. Avoid smaller models — they default to generic 'I hope this finds you well' patterns even when instructed not to.
Can I customize the Cold Email Template Engine prompt for my use case?
Yes — every Promptolis Original is designed to be customized. Key levers: The 'earned reason' must be specific enough that 95% of other senders couldn't write it. 'I read your blog' is not earned. 'Your Nov 12 post on runway math — specifically the point about gross margin being a lagging indicator — changed how I model our Q1' is earned.; If you can't name one specific earned reason, you aren't ready to send. Go do 10 more minutes of research before running the prompt.
What does it cost to use this prompt?
The prompt itself is free, MIT-licensed, with no email signup required. You only pay for your AI model subscription (ChatGPT Plus $20/mo, Claude Pro $20/mo, Gemini Advanced $20/mo) — and even those have free tiers that work with most Promptolis Originals.
How is this different from PromptBase or PromptHero?
PromptBase sells prompts in a marketplace ($2-15 each). PromptHero focuses on image-generation prompts. Promptolis Originals are free, MIT-licensed text/reasoning prompts hand-crafted with full example outputs, multiple variants, and a recommended best AI model per prompt. We don't sell anything.
Explore more Originals
Hand-crafted 2026-grade prompts that actually change how you work.
← All Promptolis Originals