The Playbook - Updated 2026

R.I.P. manually
sourcing investors.

One prompt. 5,000 matched investors. 36 meetings booked in two weeks. An $800K term sheet on the table. This is the exact playbook - every prompt, every tool, every step.

5,000
Investors Matched
36
Meetings in 2 Weeks
$800K
Term Sheet
claude - investor-sourcing.prompt live
> load business_case.md ... loaded (1.2KB) > load icp_profile.md ... loaded (0.8KB) > match investors against thesis... scanning family offices, VCs, allocators... filtering by ticket size, stage, geography... enriching with decision-maker contacts...
5,000 matching investors surfaced 2,847 direct emails verified 1,204 top-priority targets flagged
> export to smartlead + clay ... done > outreach_ready = true
The Story Behind It

4 months. 350 contacted. Zero replies.

Last month at dinner, a founder told me he'd spent four months hunting for the right investors. He'd grinded through Crunchbase, LinkedIn Sales Nav, and every VC list in his inbox. 350 contacted. Zero replies.

I opened my laptop right there at the table.

"You're doing this the hard way."

We spent 46 minutes building one Claude prompt, loaded with his business case (product, traction, financials) and his ideal investor profile (ticket size, thesis, track record).

Hit enter. Finished dinner. Came back an hour later.

5,000 matching investors. Then I dropped them into Clay, ran enrichment, and pulled decision-maker contact details for every fund. Total time: a couple hours. He started outreach that night.

Two weeks later
36
Meetings booked
$800K
Term sheet on the table
2 weeks
From dinner to diligence
The Playbook

Five steps. A couple of hours. A pipeline.

Here's the exact process. Steps 1 and 2 are previewed below - the rest unlock once you grab your copy.

01
Build your deal memo

Codify your raise into a single source-of-truth doc

Before you talk to the model, talk to yourself. A one-page deal memo covering product, traction, financials, and use of funds becomes the ground truth every downstream step compounds on.

  • Product - the one-sentence what and why-now
  • Traction - the three numbers that force a second meeting
  • Use of funds - how the round buys 18 months of compounding
Claude Notion Google Docs
02
Define your ICP

Write the investor profile you wish a database had

Ticket size, thesis overlap, stage, geography, past checks written into adjacent businesses. Treat the ICP as the query - not an afterthought - and every match downstream gets sharper.

  • Ticket band - $50K minimum, $500K sweet spot, $2M ceiling
  • Thesis signals - last 10 checks, board seats, LP composition
  • Geography - where they've led, where they'll travel, where they won't
Claude Crunchbase PitchBook
Locked
03
The mega-prompt

One prompt that turns your memo into 5,000 matches

The exact Claude prompt, including the reasoning scaffolding, deduplication logic, and output schema. Runs in ~60 minutes and hands back a structured CSV of every candidate fund with partner-level context.

Locked
04
Clay enrichment

Enrich every fund down to the decision-maker's inbox

The Clay table schema, the waterfall recipe, the validation prompts, and the exact columns that matter - built to hit 90%+ deliverability without paying for Apollo at enterprise tier.

Locked
05
The outreach sequence

A 3-touch sequence built for 25%+ reply rates

Subject lines, P.S. social proof patterns, soft CTAs, and the exact cadence. The version we deploy across Smartlead right now - with the metrics to back it.

The Claude mega-prompt

Plugs in your memo, spits out 5,000 ranked candidates

# role You are a capital-markets analyst at a top placement agent. You map founder memos to the investors most likely to lead a round. # inputs business_case: <inserted below> icp: <inserted below> # task For each candidate investor, return: - firm_name, fund_size, ticket_band - thesis_overlap_score (0-100) - last_3_checks, partner_best_fit - warm_intro_path_rank, contact_priority

Full prompt - including the reasoning scaffold, output schema, and failure-mode guardrails.

The Clay waterfall recipe

Verified decision-maker emails without enterprise spend

# columns firm_name ← from Claude website ← Clay website find people_at_firm ← LinkedIn scrape partner_filter ← role = Partner|GP|MD # waterfall try: email_from_linkedin fallback: apollo_email_lookup fallback: dropcontact_ai_pattern validate: zerobounce (deliverable only)

Full recipe - the Clay table template, credit math, and the validation stack.

Instant Access

Send me the playbook

The Claude mega-prompt, the Clay recipe, the outreach sequence, and the exact metrics - all on the next page.

Mega-prompt
Clay recipe
Outreach sequence

Your information is confidential. We may reach out to discuss how Artane can support your raise.

Playbook unlocked - bookmark this page

Everything below is yours. Copy the prompts, clone the recipes, ship the outreach.

The AI Investor Sourcing Playbook

Five steps. A couple of hours. Every prompt, every tool, every number. Start to finish, this is what we ran to turn one founder's dinner-table panic into an $800K term sheet in two weeks.

01
Foundation

Codify your raise into a one-page deal memo

Every downstream step compounds on the clarity of this document. Think of it as the system prompt for your raise - if it's vague here, every candidate list, every email, every pitch gets vague too.

The template

deal-memo.md
# Deal Memo - [Company]

What we do      one sentence. no jargon. no "platform".
Why now          the unlock that makes this a 2026 bet, not a 2022 one.
Traction         three numbers. revenue, growth, or a usage metric that proves pull.
Team             the reason you, specifically, win this market.
Raise            size, structure, close date, lead wanted.
Use of funds     how the round buys 18 months of compounding.
Milestones       what does a 3x markup look like in 12 months?
Comparables      two public or private benchmarks that anchor the thesis.

Rules that actually matter

  • One page. Not two. If it's two, you don't understand your raise yet.
  • Numbers, not adjectives. "$142K MRR" beats "strong momentum" every time.
  • Name the lead you want - stage, check size, thesis. The model will use it.
02
Targeting

Write the ICP you wish a database had

Most founders describe their investor target in two lines: "Seed VC, US or Europe, writes $500K-$2M." That's not an ICP - that's a filter on Crunchbase. Write the actual shape of a fit.

icp.md
# Ideal Investor Profile

stage               pre-seed, seed, or seed extension only
ticket_min          $50,000
ticket_sweet_spot   $500,000
ticket_max          $2,000,000
check_history       has led at least 2 of last 10 deals in adjacent sectors
thesis_signals      public posts, portfolio overlap, LP base profile
geography           US (Tier 1), EU, MENA family offices welcome
will_travel         Dublin, London, Lisbon for diligence
anti_patterns       no tourist investors, no index funds, no platform-only shops
board_stance       prefer observer, not board seat at seed
Do this

Name 3 funds you'd take a check from tomorrow. Reverse-engineer the signals. That's your ICP.

Skip this

Generic "B2B SaaS VCs." Every fund writes B2B SaaS. You're filtering on noise, not signal.

03
Matching engine

The Claude mega-prompt

This is the one. Paste your deal memo and ICP where indicated, hit run, and walk away for an hour. The model scans public data, portfolio overlap, and fund signals, then returns a ranked list with partner-level fit.

mega-prompt.md
# Role
You are a senior capital-markets analyst at a top-tier placement
agent. You have 20 years of experience mapping founder memos to
the investors most likely to lead a round. You do not hallucinate
funds - if you are not certain a fund exists, you exclude it.

# Inputs
<business_case>
{PASTE_DEAL_MEMO_HERE}
</business_case>

<icp>
{PASTE_ICP_HERE}
</icp>

# Task
Return the 5,000 investors most likely to take a first meeting.
For each, reason through:
  1. Thesis overlap with the business case (0-100)
  2. Ticket-band fit vs. their last 10 publicly known checks
  3. Stage fit vs. their recent deployments
  4. Geography fit vs. their travel + portfolio footprint
  5. Warm-intro path rank (1 = via portfolio founder, 5 = cold)

# Output schema (CSV)
firm_name, fund_size_usd, ticket_band, partner_best_fit,
partner_title, thesis_overlap_score, last_3_checks,
warm_intro_path_rank, contact_priority, notes

# Hard constraints
- Exclude any firm you cannot verify exists in public data
- Rank by (thesis_overlap_score * 0.5 + stage_fit * 0.3 + ticket_fit * 0.2)
- Flag top 500 with contact_priority = HIGH
- Do not include LPs-only shops
- Do not include tourist investors (< 2 checks in sector, last 18 mo)

# Self-check before you finish
1. Are all firms real and active in 2026?
2. Does ranking reflect fit, not brand?
3. Would a partner at a Tier 1 firm endorse this list?

How it runs in practice

StepToolWall timeOutput
Paste inputsClaude (Opus, 1M context)2 minPrompt loaded
First passClaude~45 min~5,000 ranked candidates
Dedupe & validateClaude + Clay~20 min~3,800 verified firms
ExportCSV → Clay table5 minReady for enrichment
04
Enrichment

The Clay waterfall - down to the decision-maker's inbox

Matching gives you the firms. Clay gives you the humans. The waterfall is the difference between 12% deliverability and 91% - and the difference between a Smartlead warmup that survives and one that gets your domain cooked.

Clay table schema

clay-waterfall.yaml
# Input columns (from Claude CSV)
firm_name, partner_best_fit, contact_priority

# Enrichment waterfall
1. find_company_domain       → Clay company find
2. find_people_at_company    → LinkedIn scrape, filter: Partner|GP|MD|Principal
3. best_match_partner        → fuzzy match on partner_best_fit
4. email_waterfall:
     a. email_from_linkedin    → if public
     b. apollo_lookup          → fallback
     c. dropcontact_ai_pattern → fallback (pattern inference)
     d. zerobounce_validate    → keep only "deliverable"
5. enrich_context:
     - last_3_portfolio_adds (for personalized P.S.)
     - recent_linkedin_post (hook for subject line)
     - mutual_connections (warm-intro path)

# Output
partner_name, partner_email, partner_linkedin,
recent_post_hook, mutual_connections, warm_intro_path

Credit math (so you don't torch your budget)

StepCredits / row3,800 rows
Company domain13,800
LinkedIn people scrape27,600
Email waterfall3 avg11,400
Validation13,800
Total~26,600 credits

At Clay's Explorer plan, that's roughly $149/month of credit - less than one senior analyst day.

05
Outreach

The 3-touch sequence that hits 25%+ reply rates

Everything above is wasted if your sequence sounds like a sequence. The template below is the exact one we run: max two sentences per touch, P.S. social proof, soft CTA, outcome-focused.

Touch 1 - Day 0

touch-1.email
Subject:  {{firm_name}} + {{company}}?

Hi {{first_name}},

Saw your check into {{recent_portfolio_add}} - we're
building in the same neighborhood, with {{traction_hook}}
in the last 90 days.

Worth a 20-min intro call this week?

{{founder_name}}

P.S. {{social_proof_one_liner}}

Touch 2 - Day 3

touch-2.email
Subject:  (reply in thread)

Quick one - attaching a one-pager on {{company}}.
Numbers that'll make your scout email me back on Monday.

Worth 20 minutes if the shape fits?

[one-pager.pdf]

Touch 3 - Day 7

touch-3.email
Subject:  (reply in thread)

Last nudge - closing the round in 6 weeks, prioritizing leads
who can commit a first call by end of week.

If now's not the moment, say the word and I'll circle back on
the Series A.

The numbers we see

MetricTypical cold outboundThis playbook
Deliverability60-72%91%+
Open rate18-28%54%+
Reply rate1-3%25%+
Meeting book rate0.3-0.6%4-7%
Time-to-term-sheet4-9 months2-6 weeks

Want Artane to run all five steps for you?

We do this for 6-8 companies a quarter. If you'd rather skip the setup and start the week with a warm pipeline, let's talk.

Book a Discovery Call