A fundraise is a sales process wearing a suit. Thirty to forty parallel conversations, each moving at its own pace, each one a dropped thread away from sending exactly the wrong signal.
Most of what eats a founder's time during a raise isn't the judgment calls. It's the mechanical layer underneath them: researching funds, mapping intros, drafting outreach, keeping a tracker current, writing monthly updates, following up on every "no." That layer is exactly what Claude Code is good at removing, provided you set it up properly first.
This is the full playbook: the context files that run everything, and the seven workflows built on top of them.
1. Build the context that runs your entire raise
Every prompt below pulls from three files, living in a folder called fundraise/ that you open in Claude Code at the start of every session.
company.md — the operational version of your business. What you do, who for, how you make money, your traction, and what makes you defensible. Include what you're raising for: the specific milestones this round funds.
raise.md — the parameters of this specific round: stage, target raise, valuation, close date, lead investor status, and what this round funds. Claude uses this to calibrate every research prompt and outreach message to the right stage and framing.
network.md — a structured list of the people in your network likely to have relationships with investors: strong connections, medium connections, advisors and angels, portfolio companies you know, and previous employer networks.
Run this once to scaffold the whole project:
I'm setting up a fundraising operations project. Scaffold the
full folder structure with the base files.
Create this structure:
fundraise/
├── CLAUDE.md
├── company.md
├── raise.md
├── network.md
├── target-list/
│ ├── raw-longlist.md
│ └── ranked-shortlist.md
├── intro-map/
│ └── paths.md
├── outreach/
│ ├── templates/
│ └── sent/
├── pipeline/
│ ├── tracker.md
│ └── weekly-review.md
├── data-room/
│ ├── index.md
│ ├── faq.md
│ └── materials/
└── updates/
├── active-investors.md
└── no-nurture.md
For CLAUDE.md, write the project-level context instructions:
At the start of every session:
1. Read company.md, raise.md, and network.md in full
2. Load the current state of pipeline/tracker.md
3. Confirm the current date and last session date
4. Ask: "What are we working on today?"
Rules for all outputs:
- Never produce generic content. Every output should
reference specific fund names, partner names, or
portfolio companies pulled from research.
- Flag any output that relies on assumed information
rather than confirmed data.
- Draft mode by default: no emails sent, no messages
posted without explicit approval.
- When context is ambiguous, ask before producing.Once the structure exists, paste real content into each file. The project is now the memory of your raise.
2. Find the 40 funds worth your time
The default mistake is building a target list by feel: funds you've heard of, funds mentioned at a conference, funds that backed a company you admire. Claude builds the list from your context files instead, researching by stage, vertical, and thesis alignment.
Step 1 — the longlist. This prompt reads your company and raise parameters and goes looking for 150–200 funds that could plausibly write a check:
Read company.md and raise.md.
Research the VC landscape and build a longlist of 150-200
funds that could plausibly invest in a company like ours.
Criteria for inclusion:
- Invests at [stage from raise.md]
- Check size consistent with [raise target from raise.md]
- Has publicly invested in, or written about, our category
or an adjacent one (based on company.md)
- Has made at least one investment in the last 12 months
(flag any fund quiet for 9+ months as possibly between funds)
For each fund, find: name and website; the partner most
relevant to our category; one sentence on their thesis,
quoted from a primary source with URL; their last 3
announced investments with dates; whether any portfolio
company is a direct competitor (Y/N).
Do not include funds where you cannot find evidence of
thesis alignment. Save the full table to
target-list/raw-longlist.md.Step 2 — rank it down. Once the longlist exists, run the ranking pass, weighted in this order: thesis alignment, stage and check size fit, recent activity, portfolio conflict (eliminates, not just penalizes), and warm path via network.md. Save the output — 30 to 50 funds with a target partner, a one-sentence fit rationale in the fund's own words, warm-path status, and the one open question to resolve before the meeting — to target-list/ranked-shortlist.md.
The result is a researched, ranked list built entirely from public data, without starting from a single fund name.
3. Map every warm intro before you send a cold email
A warm intro converts to a first meeting at two to five times the rate of a cold email to the same partner. The right person is a founder who already has a relationship with that fund, a portfolio founder, a co-investor, or a credible mutual who can speak to your work specifically.
Export your LinkedIn connections (Settings → Data Privacy → Download your data) and drop the CSV into fundraise/network/, along with any other network exports you have. Then run:
Read raise.md, network.md, and target-list/ranked-shortlist.md.
Also read any files in the network/ folder (LinkedIn export,
contact lists, etc.).
For each fund on the ranked shortlist, find the most likely
warm intro path from me to the target partner.
For each fund, identify:
1. BEST PATH: [my connection] → [intermediate if any] →
[partner]. Quote the evidence for the relationship.
2. RELATIONSHIP QUALITY: strong / medium / speculative.
3. THE ASK: a specific one-sentence ask to my connection
for the intro.
4. ALTERNATIVE PATH, if one exists.
For funds with no identifiable path, mark "cold" and note
what kind of connection would help. Save to
intro-map/paths.md.For funds where you're cold, the output tells you what connection to go find, via LinkedIn mutuals, event attendance, or an advisor's network.
4. Write outreach that reads like it was written for that investor specifically
The email that gets a reply references something real: a specific portfolio company, a thesis the partner articulated publicly, a recent investment in an adjacent space. Doing the research is itself a signal.
Build the dossier first, for each fund on your shortlist:
Read company.md and raise.md.
Research [fund name], specifically [partner name] as the
target for our outreach.
Find: their stated thesis (quote the most specific passage,
with source and date); 2-3 relevant portfolio companies and
why they're adjacent; any public statement in the last 12
months about a trend relevant to our space; the single most
compelling connection between our company and their thesis;
and the most likely reason they'd pass on us.
Save to outreach/templates/[fund-slug]-dossier.md.Then write the first email — 150 words maximum. Not a pitch, a request for a conversation:
Read company.md, raise.md, and
outreach/templates/[fund-slug]-dossier.md.
Write a cold outreach email to [partner] at [fund].
Rules:
- 150 words maximum, hard limit.
- Open with a specific reference to something real about
their portfolio or thesis. Not a compliment — an observation.
- One sentence on what we do, in customer language.
- One sentence on traction: the most concrete number available.
- One sentence on why we're reaching out to them specifically,
following logically from the opening.
- Close with a direct ask: a 25-minute call.
- No "great fit," no "hoping to connect," no "let me know if
you have bandwidth."
Also write: a subject line under 8 words, and a one-line
follow-up for 5 days out if no reply. Save to
outreach/templates/[fund-slug]-email.md.For a warm intro, the prompt changes: you need a personal request to your connection, plus a forwardable paragraph, 80 words maximum, written in your connection's voice, not yours.
5. Run the pipeline without losing a thread
The fundraise pipeline has one specific failure mode: lose track of a single thread — a missed follow-up, a forgotten data request — and the signal you send is exactly the wrong one.
pipeline/tracker.md is the single source of truth, with four sections: Active Conversations, Warm but Dormant, Passed, and Not Yet Contacted. Each entry: fund and partner, status, last action, next action, and notes.
Every Monday, run the weekly review:
Read company.md, raise.md, and pipeline/tracker.md.
Today is [date].
1. FLAG STALE CONVERSATIONS: any where the last action was
more than 5 days ago with a pending next action.
2. FOLLOW-UP DRAFTS: for each stale conversation, draft a
follow-up referencing something specific from the last
interaction, adding new information where available, under
80 words, closing with a specific ask. No generic
"just checking in."
3. MOMENTUM AUDIT: which conversations are close to a
decision, and what would accelerate them?
4. PIPELINE HEALTH: count by stage. Based on typical
conversion (~30 first meetings → 5 term sheets → 1 close),
is current volume enough to close [raise target] by
[close date]? If not, how many new first meetings are
needed?
Save follow-up drafts to pipeline/weekly-review.md.A separate timing prompt helps you sequence conversations deliberately: which are closest to a term sheet, which honest momentum signals you can use in outreach, which 3–4 funds to push toward parallel decisions, and which cold threads from the last 30 days are worth reactivating.
6. Build a data room that answers questions before they're asked
The standard early-stage data room, in the order an analyst reads it: company overview, pitch deck, financials, metrics, product, customers, team, legal, market, and — the document that matters most — the FAQ.
The FAQ turns the data room from an archive into a sales document. It anticipates the questions every analyst will ask and answers them honestly, including the ones with an uncomfortable answer:
Read company.md, raise.md, and any financial files available
in data-room/financials/.
You are playing the role of an investment analyst at a
top-tier VC fund reviewing our company for the first time.
Generate the 15 questions you would ask based on what you
can read in these files, and what you'd expect to find
absent for a company at our stage.
For each question: write it the way an analyst would ask it;
draft an honest, direct answer, explicit about mitigation
where the answer involves a weakness; flag any question
where the honest answer requires data we haven't provided
as "data needed."
Do not soften questions. Prioritize unit economics and path
to profitability, retention/NRR, competitive defensibility,
team gaps, key-person risk, and vertical-specific risks.Review the output, fill the "data needed" gaps yourself, and save as data-room/faq.md. It's not a public document — it's your internal preparation for every conversation that goes past the first meeting.
7. Keep every investor warm, including the ones who said no
The fundraise doesn't end at the close. It ends when the next raise starts, and the investor most likely to lead your Series A is often someone who passed on your seed, or who's been quietly following your progress for a year without being on the cap table.
The monthly update that actually gets read is under 300 words: the number that moved most, one highlight, one honest lowlight (a perfect month is suspicious), four or five metric lines, and a specific ask or nothing at all.
The no-nurture sequence keeps every fund that passed warm, on a cadence matched to how much they engaged:
Tier 1 — got to a second meeting or later. Follow up every 6–8 weeks with a one-paragraph update referencing their specific concern.
Tier 2 — one meeting, then passed. Keep on the monthly update list; send a personal note if the milestone they cited gets hit.
Tier 3 — passed without a meeting. Quarterly check-in maximum, one sentence, no ask.
Run monthly to generate the touchpoints:
Read company.md and updates/no-nurture.md (tracks every fund
that passed, tier, stated reason, and last contact date).
For each Tier 1 and Tier 2 fund due this month, write a
brief re-engagement email: one paragraph, four sentences
maximum. Lead with the metric most relevant to what they
said when they passed. Do not ask for a meeting or re-open
the conversation unless we've hit the milestone they named —
if we have, say so directly. Close low-key: "Keeping you
posted as we build."
Flag any Tier 1 fund not contacted in 90+ days as overdue.
Save drafts to updates/no-nurture-drafts-[month].md.The point of all this
None of this replaces the judgment a fundraise requires. What it removes is the thirty hours a week of mechanical work sitting underneath those decisions — the research, the drafting, the tracking. Remove that layer, and the time you have left goes to the conversations that actually move the round forward.
