TL;DR: An investor update goes to people who can fire you, so these 20 investor update prompts are organized around the actual failure modes: buried bad news, vague metrics, no ask, and drift from last month's numbers. They cover the monthly cycle end to end, plus two rules — check a vendor's own retention terms before pasting real financials, and verify every model-generated number yourself.
Two rules before the prompts, because both get skipped under deadline pressure and both matter more here than in almost any other document a founder drafts.
What could go wrong if you paste this into the wrong AI account?
The risk is that a consumer AI account may retain your conversation, or use it to train the underlying model, on terms that differ sharply from what a business or enterprise plan of the same product offers — and the difference is not always obvious from the chat window. An investor update routinely contains real MRR, runway, a live fundraise number, or a customer name. None of that belongs in a consumer AI chat until you have actually checked that specific plan's retention terms, not assumed them from a different plan or from what you remember reading before.
The practical rule that follows from all three pages: know which specific product tier you are typing into, because "ChatGPT" or "Claude" or "Gemini" is not one policy — a personal account and a business seat of the identical product can answer this question in opposite directions. If you are not certain which tier you're on, treat the chat as if it trains on your input, because that is the default posture across all three vendors' consumer tiers as of the dates above.
Why might the numbers in an AI draft be wrong?
Because a language model is built to produce a fluent, plausible next sentence, and a fabricated metric can be exactly as fluent as a real one — the model has no internal check against your actual books. In most writing this is a minor risk. In a document going to the people who fund your company, a single invented number that reaches an investor is a real, not theoretical, problem: it can misstate the state of the business to the person deciding whether to keep backing it.
This is not a defect specific to one model or one prompt style. It is a general property of how these systems are trained to hallucinate, and it deserves more than a sentence here — our deep dive on why ChatGPT makes things up covers the actual mechanism, with primary sources, rather than repeating the surface-level "it predicts tokens" explanation that stops short of being useful. The practical takeaway for the twenty prompts below is narrower and non-negotiable: every metric, date, and dollar figure in an investor update must come from your own source of truth, checked by you, before it reaches a reader who is deciding whether to keep funding you. None of the prompt templates that follow are a substitute for that check — they draft structure and prose; you supply and verify every number.
What should a monthly investor update actually cover?
A monthly investor update should answer four questions, in this order: is the month good, bad, or mixed; what moved and by how much; what is broken and what you are doing about it; and what you specifically want from the reader. Most templates in circulation list eight or ten sections. Investors reading a stack of these in one sitting only reliably read four.
| Section | What it answers | Do investors read it first? |
|---|---|---|
| Headline / one-liner | Is this month good, bad, or mixed | Yes, always |
| Metrics | What moved, and against what prior number | Skimmed, compared to last month |
| Risks / blockers | What is not working, and the plan for it | Yes, if it is present |
| The ask | What you want the reader to actually do | Yes, if they can act on it |
The prompts below are built around that shape rather than around a generic template, because a founder's real problem with investor updates is rarely "I don't know the sections." It is that the honest version of sections two and three is uncomfortable to write, and the ask gets vague because asking for something specific feels like exposing a weakness. The prompts in the bad-news and ask sections further down are the ones worth spending the most time on.
1. Full monthly update from raw notes
You are drafting a monthly investor update for [Company], a [stage] startup in [space].
Raw notes: [paste bullet notes — metrics, what happened, what didn't, what you need].
Structure the output as four sections: Headline (one sentence, the real state of the month),
Metrics (with the prior month's number next to each), Risks & Blockers (name the risk, what
caused it, what you're doing), and The Ask (one specific request).
Do not add any number, percentage, or claim that is not in my notes above.
Tone: direct, plain, no spin. Max 300 words total.
2. One-line headline generator
Given these facts about [Company] this month: [paste 3-5 bullet facts, including any bad news],
write five candidate one-sentence headlines for the top of an investor update.
Each headline must be true even if the reader stops after this one sentence.
Do not use a positive framing if the honest framing is negative or mixed.
Rank the five from most to least direct.
3. Quarterly rollup from three monthly updates
Below are my last three monthly investor updates for [Company]: [paste all three].
Write a quarterly update that does NOT just restate month three.
Identify the actual trend across the three months (accelerating, flat, or declining) for each
metric I mentioned, and write one paragraph on what changed strategically this quarter that a
single month's update wouldn't show.
Flag anywhere the three updates contradict each other before you write the summary.
4. Board version and investor version from the same notes
Using these raw notes: [paste notes], produce two versions.
Version A: a board-meeting bullet list, dense, assumes deep company context, no explanation of
acronyms or prior decisions.
Version B: a prose investor update, assumes only what a monthly-update reader would already know,
spells out anything version A left implicit.
Both versions must agree on every number and fact — flag it if you can't make them agree.
How do you turn metrics into a narrative without spinning them?
You turn metrics into a narrative by stating the number, its direction against the prior period, and the one factual reason for the change, in that order, with no adjective doing work the number should be doing. A metric narrative goes wrong in one of two directions: either it reads as a spreadsheet with no story, or it reads as marketing copy with no spreadsheet.
The safest guardrail for a model here is the same one you would use in a code review: give it the numbers and forbid it from adding any it wasn't given. A model does not know your churn was actually the loss of one large account rather than a systemic problem, so it will guess at a cause if you let it. Tell it the cause; ask it only to write the sentence.
5. Metrics table to narrative, grounded only
Metrics this month for [Company]: [paste a table — metric, this month, last month].
Cause of each significant change (fill in yourself, don't leave blank): [paste your own one-line
cause per metric].
Write one paragraph per metric: the number, the direction, and the cause I gave you.
Do not invent a cause I did not provide. Do not add an industry benchmark or comparison unless I
give you one below: [paste a benchmark only if you actually have a sourced one, otherwise leave blank].
6. Context without a fabricated benchmark
This metric moved from [old value] to [new value] this month: [metric name].
Write two sentences of context using only our own last six months of this metric: [paste the
six prior values].
Do not compare this to an industry average, competitor number, or benchmark of any kind unless I
explicitly provide one — if you don't have real context to add beyond our own trend, say so
plainly instead of inventing a comparison.
7. Plain-English cohort or retention explainer
Explain this retention/cohort data to a non-technical investor: [paste the cohort table or
retention curve description].
Two sentences max. State what percentage of a cohort is still active after the stated period, and
whether that is improving, flat, or worsening against our prior cohorts: [paste prior cohort
numbers for comparison].
No jargon — assume the reader has never seen a cohort table.
How do you deliver bad news without burying it?
You deliver bad news by putting the fact in the first sentence, before any context, cause, or plan. This is the single highest-value skill in this entire post, because it is the one founders systematically get backwards under pressure. The instinct when a number is bad is to lead with the explanation, so the reader absorbs the reasoning before the fact lands — which reads, correctly, as spin.
A useful test: read your bad-news sentence and delete every word that is doing emotional work instead of factual work — "slight," "a bit," "somewhat," "just a small." If the sentence still says the same thing after you delete them, they were softening it. If deleting them changes what the sentence claims, you were misstating the fact, not just softening its delivery.
8. The no-spin bad-news paragraph
Write a paragraph disclosing this to investors: [state the bad news fact plainly, e.g., "MRR fell
9% this month" or "we lost our largest customer"].
Sentence 1: the fact, no qualifier, no softening word.
Sentence 2: the specific cause (fill in): [your actual cause].
Sentence 3: the specific action you're taking (fill in): [your actual next step].
Do not add words like "slightly," "just," "only," or "a bit" before any number.
Do not open with the cause or the silver lining before the fact.
9. Missed-target explainer
We set a target of [target] for [metric] this month/quarter. We hit [actual result].
Write two sentences: the gap, stated as a number or percentage, then the real cause (fill in,
don't let the model guess): [your actual cause].
This is not an excuse-finding exercise — if the honest cause is "we set the target wrong," say
that instead of blaming external factors.
10. Runway and burn, stated plainly
Our runway is currently [X months] at [burn rate/month], cash on hand [$Y].
Write one sentence stating this plainly, with no reassuring framing attached ("but we're
confident," "however we expect," etc. are banned in this sentence).
If a plan changes the runway (fundraise in progress, cost cuts planned), state that as a separate,
clearly labeled second sentence — do not blend the two so the runway sounds better than it is.
11. The skeptical-investor draft audit
Here is my full investor update draft: [paste your entire draft].
Act as a skeptical existing investor reading this for the fourth time this year. List every
question this draft raises that it does not itself answer — specifically look for: a number with
no comparison point, a risk mentioned without a stated response, a metric that contradicts last
month's stated trend, and any sentence a careful reader would read as spin.
Output a numbered list of unanswered questions only. Do not rewrite the draft.
This one is worth saving and running on every update before you send it, not just when something looks off. A draft that reads fine to its author is exactly the case where a second, skeptical pass earns its keep — the questions it surfaces are close to the ones you would actually get asked in a call, at which point answering them costs you a scheduling delay instead of a sentence.
12. Corporate-speak-to-plain-English pass
Here is a paragraph from my investor update: [paste paragraph].
Rewrite it removing all hedging and corporate softening — "leverage," "circle back," "at this
juncture," "we're excited to share," passive voice that hides who did what.
Keep every fact and number exactly as stated. Flag anything you removed that might have been
covering for a fact I actually need to include, rather than just wordy phrasing.
What should the ask actually say?
The ask should name one specific thing, addressed to a type of person or resource you actually need, with a clear description of what a useful reply looks like. "Let me know if you can help" gets no replies, because the reader has no idea what action would satisfy it. A specific ask trades the appearance of humility for an actual response.
Most updates either skip the ask or list three or four of them. Both fail for the same reason: a reader who has to choose which of your asks matters most usually chooses none of them. Pick the one ask that would move the needle most this month, and cut the rest to a single follow-up sentence at most.
13. Specific ask generator
I need help with: [vague version of what you need, e.g., "hiring" or "intros"].
Turn this into one specific, actionable ask: name the exact role, company type, or resource, and
describe what a useful reply from an investor looks like in one sentence.
Do not produce a general "let me know if you can help" version — if my input was vague, ask me one
clarifying question instead of guessing.
14. Ask prioritizer
Here are the things I could ask investors for this month: [list up to 5 possible asks].
Pick the single highest-value one for this specific month, given this context: [state your
current priority — e.g., closing the round, a key hire, a specific customer intro].
Justify in one sentence why this is the one ask and not the others.
Write the final one-ask version I should actually send.
How do you keep this month consistent with last month?
You keep an update consistent by checking it against the last one before you send it, specifically for silently changed metric definitions, a blocker described as solved that has reappeared, and a number whose trend contradicts what you said a month ago. Investors who read every update notice these gaps faster than a founder writing under deadline pressure does, and a caught inconsistency costs more trust than the original bad news would have.
15. Cross-update consistency check
Last month's update: [paste it].
This month's draft: [paste it].
Compare them and flag: any metric where the direction contradicts what was implied last month, any
blocker marked resolved last month that appears again this month, and any claim this month that
last month's numbers would not support.
List only the contradictions. Do not comment on writing style.
16. Metric-definition drift check
I define [metric name, e.g., "active users"] as: [your current definition].
Here's how I described this same metric in past updates: [paste 2-3 prior mentions].
Check whether the definition has silently shifted between updates (e.g., a stricter or looser
activity window, a different denominator). If it has, tell me exactly which update changed it and
how, so I can flag the change transparently instead of letting the number look better by accident.
How do you tighten a bloated draft?
You tighten a bloated draft by cutting to a target word count while treating every number, name, and date as non-negotiable, then removing hedging language as a separate pass. Length is one of the most common, least discussed reasons an update goes unread: investors receive many of these in the same week, and the ones that get opened next month are the ones that took under two minutes this month.
17. Cut to word count, facts protected
Here is my draft: [paste draft]. Target length: [target word count, e.g., 250].
Cut this to the target length. Do not remove any number, date, name, or specific claim — cut
only connective tissue, repeated context, and sentences that restate something already said.
List anything you had to cut that was a fact rather than filler, so I can decide whether to keep
it elsewhere.
18. Kill-the-filler pass
Here is my draft: [paste draft].
Identify every sentence using hedging language ("we believe," "it's worth noting," "generally
speaking") or passive voice that obscures who did something.
Rewrite each flagged sentence in direct, active voice. Do not change any fact, number, or claim —
only the phrasing.
What changes in an annual or fundraise-adjacent update?
An annual update needs the trend line across twelve months, not another month's snapshot, and a fundraise-adjacent update needs the same monthly structure recast around proof points and a specific new ask: the round itself. Both are higher stakes than a routine monthly send, because an annual letter often goes to prospective investors who have never seen your monthly cadence, and a fundraise update is being read for evidence, not just kept current.
19. Annual retrospective from twelve monthly updates
Here are my twelve monthly investor updates from the past year: [paste all twelve, or summarize
each in 2-3 bullets if the full text is too long].
Write an annual update that identifies the real throughline across the year for each key metric
(not a month-by-month recap), the two or three moments that mattered most, and how the company's
biggest risk at the start of the year compares to its biggest risk now.
Flag anywhere the twelve updates contradict each other on a number or a stated plan.
20. Fundraise-mode update recast
Here is my standard monthly update: [paste it].
Recast it for investors actively evaluating a new round: lead with the two or three strongest
proof points from the metrics section, move the ask to "we are raising [amount] at [stage/terms],
here's what closes it," and keep the same honesty standard on risks — a fundraise update that hides
a known risk is worse than one that surfaces it with a plan attached.
Do not inflate any number beyond what's in my original draft.
How does Prompt Architects fit this workflow?
These 20 prompts work as one-off pastes into any AI chat. They work better saved once, with your company's [Company], [Stage], and [KeyMetric] stored as reusable variables so you fill in only this month's actual numbers instead of retyping your context every time. That is the same system our founder AI workflow guide walks through end to end, and the same variable-based structure our 40 AI prompts for startup founders post uses across fundraising, hiring, GTM, and ops. If you are setting up that library for the first time, the founder's prompt system guide covers the migration from scattered notes to something reusable.
Stop rewriting prompts. Start shipping.
Works with ChatGPT, Claude, Gemini, Grok, Midjourney, Ideogram, Veo3 & Kling. 5.0★ on the Chrome Web Store.
Create An AccountSave the four or five prompts above you will actually run every month — the skeptical-investor audit and the bad-news paragraph are the two worth keeping closest at hand — and fill in this month's real numbers before you send anything. The prompt gets you a structured, honest-shaped draft in minutes. Making sure every number in it is true is still the founder's job, not the model's.
By Nafiul Hasan — Founder of Prompt Architects.