AI Prompts for Business Plans

AI prompts for business plan work when you draft one section at a time from a fact pack you control: problem, market, model, then financials, each with accuracy fences that block invented customers, market size, and revenue. You leave with ROLE / TASK / FORMAT shapes for GPT-5.5 Instant, GPT-5.6, Claude Sonnet 5, or Gemini 3.5 Flash, plus audit prompts that mark gaps with [NEED FACT]. This guide covers plan drafting for founders and operators. It skips ops SOPs in the ChatGPT prompts for business guide. Paste customer notes, pricing you own, and real cost lines first. Soft tip: PromptMake /text can turn a rough plan ask into a structured scaffold at https://promptmake.net/text before you paste into your chat model.

Who AI prompts for business plan help

You need a clear written plan and want the model to turn scattered notes into section drafts without inventing TAM figures, logos, or break-even dates. The patterns fit solo founders who write a lean plan for co-founders or early hires, operators who refresh an existing plan before a board meeting, and small-business owners who need a bank or grant narrative grounded in facts they can defend. Advisors who help clients structure problem and market sections land here too.

Skip this page if you want internal SOPs, weekly status email, or decision memos for ops teams; that job lives in the ChatGPT prompts for business guide. Skip it if you want sales outreach or CRM pipeline notes; those jobs live in the AI prompts for sales guide. This article covers plan writing: what problem you solve, who buys, how you make money, and which numbers you may state.

Treat the model as a section editor with constraints. You supply FACT_PACK, section goal, audience, and banned claims. The model proposes structure, clearer verbs, and consistent headers. You reject anything that invents a customer count, a market dollar figure, or a margin your spreadsheet never showed.

Core prompt pattern for business plan sections

Strong AI prompts for business plan give four inputs before any tone request: the fact pack, the section name, the reader, and the accuracy fence. The fact pack holds customer interviews, pricing you may state, unit costs you measured, competitor names you verified, and numbers marked uncertain. The section name is problem, market, model, financials, team, or go-to-market. The reader is the person who must act: co-founder, lender, grant reviewer, or board member. The accuracy fence forbids invented TAM, fake logos, made-up churn, and revenue lines you did not approve.

Paste those four blocks near the top. Put format rules next. Put banned phrases at the end so they survive a long paste. GPT-5.5 Instant, GPT-5.6, Claude Sonnet 5, and Gemini 3.5 Flash all follow labeled blocks well. Vague "write my business plan" prompts produce generic startup essays because the model has no fact pool and no forbidden list.

Aim for one section per thread, or one section per message in a short thread. Mixing problem discovery and five-year P&L in the same pass muddies length and invents numbers to fill empty cells. Keep FACT_PACK outside the chat in plain text. Update it when pricing, costs, or customer evidence changes. Feed it into every prompt that touches plan language.

Role, task, format skeleton

Copy this skeleton and fill the brackets with your material:

ROLE: You are a business plan writing assistant for [venture or company]. You draft from FACT_PACK only. You never invent customers, logos, market size, growth rates, margins, funding amounts, or timelines.

TASK: Turn FACT_PACK into the [problem | market | business model | financials | team | GTM] section for [reader]. Keep every claim tied to FACT_PACK. Mark gaps with [NEED FACT].

FORMAT: Use the section headers listed under OUTPUT_SHAPE. Short sentences. Numbers only when present in FACT_PACK. Bullets preferred for lists. Cap prose sections at the word limits you set.

READER: [role and what they must decide after reading]

FACT_PACK: [paste customer notes, pricing, costs, competitor list, traction metrics, and uncertain items marked with ?]

OUTPUT_SHAPE: [list required subsections for this plan section]

RULES: No hype adjectives. Ban words like revolutionary, disruptive, and guaranteed. If a figure is missing, write [NEED FACT] instead of guessing. Never invent TAM, SAM, SOM, or CAGR.

REMINDER: Never invent customers, market dollars, or revenue. Use [NEED FACT] for gaps.

That reminder line stops confident fake market math. Models love tidy TAM pyramids. Your rule forces a gap marker you can fill from a real source or leave blank until research lands.

Accuracy fences and section packs

Build a short ACCURACY_FENCE block you paste under RULES every time:

ACCURACY_FENCE: Forbid new customers, logos, case studies, TAM/SAM/SOM, CAGR, ARR/MRR, margins, CAC, LTV, burn, runway, and close dates unless they appear in FACT_PACK. Quote ranges only when FACT_PACK states ranges. Prefer "unknown" over a confident guess.

Problem pack fields: who hurts, what they try today, cost of the pain in their words, evidence source (interview date or ticket ID). Market pack fields: buyer segment, geography you serve, competitors you named, pricing bands you observed, sources for any size claim. Model pack fields: offer, price, who pays, delivery steps, key partners. Financials pack fields: unit price, unit cost, fixed costs you measured, current revenue if any, assumptions labeled ASSUMPTION.

When a section needs a figure you lack, tell the model to leave a labeled blank rather than invent. Example rule: "If market size is missing, write Market size: [NEED FACT: source] and continue with qualitative buyer description only." That keeps the draft useful for co-founders while protecting lender-facing numbers.

Section-by-section AI prompts for business plan

Draft in order: problem, market, model, financials. Early sections feed later ones with language and constraints. A clean problem statement gives the market section a buyer to describe. A clear model gives financials price and cost lines to use. Jumping straight to a five-year spreadsheet invites the model to invent growth curves that your evidence never supported.

Reuse the same FACT_PACK and ACCURACY_FENCE. Change only TASK, OUTPUT_SHAPE, and any section-specific rules. After each section draft, run a short audit before you move on. Carry forward only claims that survived the audit. Soft tip: if your rough ask is messy, shape the section prompt once at https://promptmake.net/text, then fill FACT_PACK yourself.

Problem and market section prompts

Problem pattern:

TASK: Draft the Problem section under 250 words for [reader]. Sections: Who feels the pain, What they do today, Cost of the status quo (only if in FACT_PACK), Why now (only if in FACT_PACK). Use customer language from FACT_PACK. Ban invented quotes. If no interview notes exist, write [NEED FACT: customer evidence] under Who feels the pain and stop inventing personas.

Market pattern:

TASK: Draft the Market section. Sections: Target buyer, Segment boundaries, Competitors named in FACT_PACK, Differentiation claims tied to FACT_PACK only, Market size. For Market size: if FACT_PACK lacks a sourced number, write [NEED FACT: market size source] and describe the buyer in plain words without dollar claims. Never invent TAM, SAM, or SOM. Cap at 350 words unless READER needs a longer grant narrative.

Second-pass market audit: "TASK: Audit MARKET_DRAFT against FACT_PACK. List every dollar figure, growth rate, and competitor claim. Reply Keep, Soften, or Remove. Soften means the claim overreaches the sources. Propose safer wording. Never add new market math."

Model and financials section prompts

Business model pattern:

TASK: Draft the Business model section. Sections: Offer, Who pays, Pricing (from FACT_PACK only), How delivery works, Key partners if named. Under 300 words. If pricing is undecided, list options labeled ASSUMPTION and mark a decision needed. Do not invent discount ladders or enterprise tiers missing from FACT_PACK.

Financials pattern:

TASK: Draft the Financials narrative (not a full spreadsheet). Sections: Unit economics from FACT_PACK, Fixed costs listed, Revenue to date if any, Near-term forecast assumptions labeled ASSUMPTION, Risks and sensitivities. Rules: every number must cite FACT_PACK or ASSUMPTION. Never invent ARR, burn, or runway. If a cell is empty, write [NEED FACT]. Prefer a 12-month view unless FACT_PACK states a longer horizon.

Financials audit pattern: "TASK: Compare FINANCIALS_DRAFT to FACT_PACK line by line. Flag any margin, growth rate, or customer count not present in FACT_PACK. Rewrite flagged lines with [NEED FACT] or remove them. Do not replace gaps with industry averages."

Optional team / GTM add-ons use the same skeleton. Team: roles filled, roles open, relevant proof from FACT_PACK only. GTM: channels you already use or plan with named owners; ban invented partnership logos.

Step-by-step business plan prompt workflow

Use one chat thread per plan version or per major reader: co-founder lean plan, lender packet, grant narrative. Dumping every audience into one long thread blurs tone and invents shared numbers that none of the readers should see. The loop below keeps a stable FACT_PACK while you swap only the section TASK and the reader. You spend free ChatGPT, Claude, Gemini, or PromptMake runs on structure, then human time on number checks and source links.

Keep FACT_PACK in a plain text file or sheet: customer evidence, pricing, costs, competitors, traction, and open questions. Mark uncertain items with ?. Update the pack when a real interview or cost quote arrives. The model should never be your market research database or your accounting system.

Step 1: Build the fact pack and fence

List what you may say and what you may not. Example: "Product: RouteDesk. Buyer: ops leads at 20 to 200 person field teams. Evidence: 6 interviews (dates listed). Price: $79/user/mo list. Cost: $12/user infra estimate (ASSUMPTION). Competitors: ToolA, ToolB. Ban: invented logos, TAM dollars, "guaranteed payback."" Ugly notes beat polished fiction.

Mark every growth rate and market dollar as either sourced or forbidden. In the prompt, tell the model to keep unknowns as [NEED FACT]. Guessing "$4B TAM" when you mean "we talked to 12 buyers in one metro" creates false investor confidence and a weak plan review later.

Step 2: Outline one section, then draft

Name the section and reader before you paste long notes. Ask for an outline first when the section is new:

TASK: From SECTION_GOAL and READER, propose an outline with headers only. Do not draft body text yet. Flag any header that needs facts missing from FACT_PACK.

Use that outline to decide what to paste next. Partial facts get [NEED FACT] subsections. Missing financial decisions stay out of the draft until a human decides them. Then run the full section skeleton with OUTPUT_SHAPE matching the outline you approved.

Step 3: Audit claims, then lock the section

Take the draft into a second message:

TASK: Audit DRAFT against FACT_PACK and ACCURACY_FENCE. For each claim about customers, market size, pricing, margin, growth, funding, or timeline, reply Keep, Soften, or Remove. Soften means the claim overreaches the sources. Propose a safer rewrite for Soften and Remove lines. Never add new commitments or industry averages.

Optional scaffold: open https://promptmake.net/text, describe "business plan section prompts for problem, market, model, and financials with accuracy fences and [NEED FACT] gaps," generate once, then paste your FACT_PACK into the returned structure. Guests get about three runs per day; free accounts get about five. Use a run to shape the prompt, then finish in your chat model.

Save the winning prompt next to the section name and model label so the next plan version reuses the same wrapper. Lock audited text into your doc before you start the next section so invented numbers cannot leak backward.

Mistakes that wreck business plan drafts

Mistake 1: Asking the model to "write a full business plan" with no FACT_PACK. The model invents market size, personas, and tidy financials that fail under the first hard question. Paste controlled notes instead.

Mistake 2: Allowing invented TAM, SAM, SOM, or CAGR. If you did not supply a sourced figure, ban market math in FORMAT. Fake pyramids create diligence debt.

Mistake 3: Mixing ops SOP tone or sales outreach tone into the plan. Ban words like revolutionary, game-changing, and guaranteed in RULES. Plan readers need evidence and clear offers.

Mistake 4: Using GPT-5.5 Instant or Gemini 3.5 Flash for the final audit on financials that commit revenue or runway claims. Fast models fit first drafts. Route the audit pass to GPT-5.6, Claude Opus 5, or Gemini 3.1 Pro when numbers go to lenders or boards.

Mistake 5: Pasting full payroll, bank balances, or personal customer PII into a consumer chat without your company's AI policy check. Redact names and account numbers. Keep sensitive fields human-typed offline.

Mistake 6: One mega-prompt that asks for problem, market, model, financials, and pitch deck copy together. Split sections. Reuse FACT_PACK; change TASK and FORMAT.

Mistake 7: Trusting the model on legal, tax, or securities wording. Ask for plain-language drafts of the business narrative. Send regulated disclosures to counsel before you file or circulate.

Mistake 8: Skipping the audit after financials. Narrative summaries hide invented margins. Prompt for Keep / Soften / Remove against FACT_PACK every time.

Model notes for business plan writing (mid-2026)

ChatGPT often defaults to GPT-5.5 Instant for fast chat. Instant fits brainstorming section outlines, drafting problem and model prose from FACT_PACK, and first-pass market narratives when you already banned invented TAM. Keep prompts short: ROLE, TASK, FORMAT, FACT_PACK, RULES. Skip long chain-of-thought slogans.

GPT-5.6 (Sol in API naming as of mid-2026) fits harder edit passes: claim audits against FACT_PACK, contradiction checks between model pricing and financials, and lender-facing drafts that must not invent runway. Give goal, constraints, and format. Drop "think step by step" padding on reasoning-class models.

Claude Sonnet 5 handles long interview paste and tidy section formats well for problem and market drafts. Claude Opus 5 fits enterprise audits when the plan must not invent options or market math. Claude Fable 5 is the widely released top tier when your workspace offers it; check your plan. Haiku 4.5 fits short header variants when latency matters more than deep audit.

Gemini 3.5 Flash fits volume work: many outline passes, section rewrites from updated FACT_PACK, and first-pass GTM drafts. Gemini 3.1 Pro fits hard reasoning over long context when you paste thick interview notes and need contradiction flags. Hedge on exact menu names in each vendor UI. They shift. Re-check the model picker when you open a new thread.

Prompting split that holds: Instant and Flash get RTF plus a short sample of your house plan voice when you need matching tone. GPT-5.6, Opus 5, and Gemini 3.1 Pro get goal + constraints + format, with an explicit "never invent" rule and a [NEED FACT] token. All need your FACT_PACK in the message. None replace a human number check before you send the plan to a lender, grant desk, or board.

Build business plan prompts with PromptMake

Write the rough ask in plain words: section name, reader, accuracy fence, and whether you need a claim audit. Open https://promptmake.net/text and generate a structured prompt once. Expect labeled sections you can fill with FACT_PACK.

Edit prices, costs, customer counts, and market claims yourself. PromptMake cannot know your spreadsheet. Paste the filled prompt into Instant or Flash for drafts, or GPT-5.6 / Opus 5 / Gemini 3.1 Pro for audits. Keep free-tier runs for scaffolding, not five synonym retries of the same weak ask.

Workflow that sticks: FACT_PACK → PromptMake scaffold → fill section facts → fast model draft → reasoning model claim audit → human number review → lock section → next section. Store one template per plan section so you do not rewrite ROLE and RULES from scratch each version.

FAQ

What are the best AI prompts for business plan drafting in 2026?

The best AI prompts for business plan drafting lead with ROLE and accuracy fences, paste a FACT_PACK, then demand FORMAT for one section at a time: problem, market, model, or financials. Add a second audit prompt that marks Keep, Soften, or Remove against your sources. Match GPT-5.5 Instant or Gemini 3.5 Flash for drafts and GPT-5.6, Claude Opus 5, or Gemini 3.1 Pro for the audit when numbers go to lenders or boards.

Can AI write a full business plan from scratch?

The model can draft structure and wording from notes you supply. It should not invent customers, market size, or revenue. Start from a FACT_PACK and messy evidence you wrote offline. Treat blank-slate "write my business plan" prompts as high risk for fiction that fails under the first diligence question.

How do I prompt AI for business plan financials without fake numbers?

Paste unit price, unit cost, and fixed costs you measured. Demand a narrative with ASSUMPTION labels and [NEED FACT] for empty cells. Ban invented ARR, burn, and runway. Run Instant or Flash for a first shape, then a reasoning audit that removes any figure missing from FACT_PACK before you share the plan.

Should I use GPT-5.5 Instant or GPT-5.6 for business plan prompts?

Use Instant for outlines, problem drafts, and first-pass model sections. Use GPT-5.6 when you need a careful claim audit, conflict checks between pricing and financials, or a lender-facing draft that must not invent market math. Run the same facts through both only when you measure quality for a recurring plan update.

How do I stop AI from inventing TAM or customer logos?

State the ban in ROLE and again in an ACCURACY_FENCE plus REMINDER line. Forbid new TAM/SAM/SOM, CAGR, logos, case studies, and revenue figures. Require [NEED FACT] when a stronger claim wants a detail you did not give. Follow with an audit prompt that compares DRAFT to FACT_PACK line by line.

Are AI prompts for business plan the same as ChatGPT prompts for business ops?

Business plan prompts target problem, market, model, and financials sections with investor or lender readers and accuracy fences on market math. ChatGPT prompts for business ops target SOPs, internal status, and decision memos. Keep those libraries separate so plan drafts stay tied to evidence and approved numbers.

Can PromptMake help with AI prompts for business plan free?

Yes. PromptMake /text turns a rough plan ask into a labeled prompt you can aim at ChatGPT, Claude, or Gemini. Guests get about three generations per day; registered free users get about five. Fill in your own FACT_PACK and section goals, then paste into Instant, Flash, or a reasoning model for the audit.