Raising money is mostly two documents on repeat. The deck that gets you the first meeting, and the monthly update that keeps everyone who said no warm and everyone who said yes from panicking. Founders pour weeks into the first and neglect the second. That's backwards. The update is what compounds. Below is how to write both with Claude, what it's genuinely good at, and where an experienced investor will smell the machine from across the table.
Quick answer: Start with a hardened narrative—problem, why now, what you've built, proof, market, plan, team, ask—then let Claude structure the skeleton and build slides in Claude Design. Use Claude for drafting and hole-finding; put the real numbers and specifics in yourself. Send monthly updates on cadence with honest lowlights and actual metrics, not template prose.
Start with the narrative, not the slides
The most common mistake I see is opening a deck template and trying to fill boxes. You end up with twelve slides that each sound reasonable and a story that goes nowhere. Investors don't fund slides. They fund a sentence they can repeat to their partners on Monday: "It's the X that does Y for Z, and the reason now is W."
So before any design, get the narrative straight. A structure that holds up: problem, why now, what you've built, proof it's working, the market underneath it, the plan, the team, the ask. Look at the order. "Why now" comes second, not buried at slide nine, because timing is the thing a sceptical investor is quietly testing the whole way through. If your business could have existed five years ago and didn't, you've got a "why now" problem and no amount of design hides it.
Claude is good at pressure-testing this. Paste your rough narrative in and ask it to play a sceptical seed investor. Not to rewrite it. To interrogate it. Where's the logical jump? Which claim needs a number behind it? What's the obvious objection you've skipped? You'll get a list of holes. That list is worth more than a polished draft, because every hole is a question you'll actually get asked in the room. Fix the holes first. Then write.
When you draft, give Claude your real material: the messy doc, the customer quotes, the actual numbers. Ask for a slide-by-slide outline, one headline claim per slide plus the two or three supporting points, no prose. The headline matters more than people think. "Traction" is not a headline. "Revenue tripled in two quarters with no paid spend" is a headline. A slide whose title is a noun is a slide doing no work.
Where Claude Design changes the deck
Until recently, Claude could write your narrative but you still had to build the actual thing in Keynote or hand it to a designer. That gap closed with Claude Design, an Anthropic Labs product launched in April 2026 that reportedly crossed a million users in its first week, with a sizeable update on 17 June (sources: anthropic.com/news/claude-design-anthropic-labs and venturebeat.com). It's a research preview on the Pro, Max, Team and Enterprise plans, powered by Opus 4.7.
The part that matters for fundraising: you describe what you want and it builds a first version. A deck, a one-pager, a prototype. Then you refine by conversation, by leaving inline comments, or by dragging things around on the canvas. Hand it your existing brand files or point it at your codebase and it builds a design system from them, so later slides pick up your colours, type and components automatically instead of you fiddling with each one. For a founder, that's the difference between a deck that looks like a product and one that looks like a template someone downloaded.
A workflow that works. Write and harden the narrative in a Claude conversation, then move it into Claude Design and ask it to lay out the deck. Feed it real screenshots of your product. There's a web capture tool that grabs elements off your live site, so a prototype slide matches the real thing. Refine on canvas. When it's right, export. Export options include PDF and PPTX, plus Canva, a folder, a shareable org URL, or standalone HTML (the June update added more destinations). PDF is what you send investors. PPTX is what you keep if you want to tinker in PowerPoint later.
Two honest limits. There's no real-time multiplayer, so you and your co-founder can't both edit live the way you would in Figma, and the comment-based collaboration gets clunky if more than one person is weighing in. And the on-canvas control, while improved in June, isn't pixel-perfect WYSIWYG. If you're the kind of founder who needs that button moved three pixels left, you'll be slightly frustrated. For a fundraise that's usually fine. You want a clear, on-brand deck fast, not a design diploma.
What investors can smell
This is the part nobody wants to say out loud. Experienced investors read hundreds of decks and updates a year. They've got a finely tuned nose for AI filler, and a deck that reeks of it reads as a founder who didn't care enough to think.
The tells are consistent. Hollow intensifiers: "revolutionary," "seamless," "game-changing." Claims with no number attached. A market section that's all top-down ("if we capture 1% of a $50bn market") and no bottom-up logic. Three adjectives where one would do. Paragraphs that are all the same length, that strange uniform rhythm AI defaults to. And vagueness where there should be specifics: "strong early traction" instead of "84 paying customers, £31 average monthly, 4% monthly churn."
The fix isn't to avoid Claude. It's to use it for the right job. Let it structure, draft, find the holes, tighten the headlines. Then you put the specifics in. The real numbers. The actual customer who said the thing. The genuine reason you started this. The judgement and the truth are yours; the drafting and formatting are the machine's. A useful final pass: ask Claude to flag any sentence that makes a claim without evidence, then go and either add the evidence or cut the claim. That one habit removes most of what makes a deck smell synthetic.
And read it aloud before you send. If a sentence sounds like a press release, it is one, and an investor will treat it as one.
The update you're probably neglecting
This is the document that actually moves money over time. Investors fund founders they've watched execute, and the monthly update is how they watch. The ones who passed at seed often lead the Series A, because eleven clean updates taught them you do what you say. Skip the updates and you're cold-emailing the same people two years later as a stranger.
Pick a cadence and never break it. Monthly is the standard. Consistency matters more than length. An update that's three weeks late signals problems louder than any bad metric. Keep the structure identical every time so readers can scan: a one-line TL;DR, the headline metrics (same metrics each month so the trend is legible), wins, lowlights, and your asks. The asks line is the one founders drop, and it's the most valuable. Investors want to help, and "we need two intros to fintech compliance leads" is a thing they can actually do this week.
Be honest in the lowlights. A polished update with no bad news reads as either naive or evasive, and sophisticated investors discount it. The founders who write candidly ("churn ticked up, here's the diagnosis, here's what we're testing") build the trust that gets the next cheque written fast.
Claude fits this well because the update is structured and recurring. Give it last month's update and this month's raw numbers and it produces a consistent draft that continues the story rather than reinventing it. You add the honest colour, the thing that happened, the call you're worried about, and send. Ten minutes, not an afternoon.
Get ahead of the FAQ
Every raise has a predictable interrogation. Unit economics, CAC and payback, why this team, what breaks at scale, who else is in. Write those answers down before you're asked. Have Claude generate the likely question list for your stage and sector, draft your answers, then sharpen each one to a tight, evidence-backed paragraph. Drop the strongest ones into an appendix after your closing slide. When an investor asks the obvious question and you flip to a slide that already answers it, you look like someone who's thought this through. Because you have.
Where to take this
If you'd rather not assemble the prompts yourself, the Fundraising pillar on Skill Locker packages this into skills built for the job: a pitch-deck-narrative skill that enforces the structure above, an investor-update-writer for the monthly cadence, an investor one-pager, a fundraising FAQ builder, and a due-diligence prep pack. They encode the craft so you bring the context and the truth. It's my own product, and there are free skills to test the quality before you pay.
Browse the Fundraising stack →
The deck gets the meeting. The updates get the cheque. Use the machine for the structure and the speed, keep the numbers and the honesty human, and you'll sound like a founder worth backing rather than a content mill that found a template.
FAQ
Why should I write the narrative before designing slides?
Most founders open a template and fill boxes, ending up with slides that sound reasonable individually but tell no story. Investors fund a sentence they can repeat to partners—"It's the X that does Y for Z, because W." Claude is good at pressure-testing a rough narrative to find logical jumps and missing evidence before you waste time polishing.
What's the difference between using Claude alone and Claude Design?
Claude writes the narrative and outlines the deck structure. Claude Design builds the actual thing—it reads your description and produces a first draft, lets you refine on canvas by commenting or dragging, pulls your brand colours from files or codebases, and exports to PDF, PPTX, Canva, or HTML. You get a deck that looks like a product, not a downloaded template.
How do experienced investors spot an AI-written deck?
They smell hollow words (revolutionary, seamless, game-changing), claims without numbers, all top-down market sizing with no bottom-up logic, uniform paragraph lengths, and vague abstractions instead of specifics. "Strong early traction" reads as synthetic; "84 paying customers, £31 average monthly, 4% monthly churn" reads as real.
What should I actually do with Claude when writing a pitch deck?
Let it structure the narrative, draft from your real material, find holes in your logic, and tighten slide headlines. Then you put the specifics in—real numbers, actual customer quotes, genuine reasons you started this. Ask Claude to flag any claim without evidence, then either add the evidence or cut the claim. The judgment and truth are yours; the drafting is the machine's.
Why is the monthly investor update more important than the pitch deck?
The deck gets the first meeting. The updates compound over time because they show investors you do what you say. Founders who passed at seed often lead Series A because clean monthly updates prove execution. Skip updates and you're cold-emailing strangers two years later instead of having warm, watching relationships.
How should I structure a monthly investor update?
Keep it identical each month: one-line TL;DR, headline metrics (same ones each month so trends are legible), wins, lowlights, and asks. Be honest in the lowlights—a polished update with no bad news looks naive or evasive to sophisticated investors. The asks line is the most valuable: "we need two intros to fintech compliance leads" is something investors can actually do this week.