AI Prompts for Private Equity: A Workflow Library
Dr. Leigh Coney
Founder, WorkWise Solutions
May 25, 2026
17 min read
TLDR: Good prompts are the difference between AI that saves a PE team hours and AI that wastes them. This library collects practical, tested prompts across the deal lifecycle: sourcing and screening, due diligence, financial analysis, IC memos, portfolio monitoring, and investor relations. The pattern behind every good PE prompt is the same: give the model a role, the context, the format you want, and the document itself, then verify the output. One rule overrides them all: never paste confidential deal data into a consumer AI account. This guide gives you the prompts and the discipline to use them.
Table of Contents
1. Why a Prompt Library Beats Ad-Hoc Prompting
Most people use AI like a search box: a vague question, a generic answer, a shrug. The same model, given a precise prompt, produces something a PE analyst can actually use. The gap between those two outcomes is prompting, and it is a learnable skill.
A firm that leaves prompting to each individual gets wildly uneven results. One associate gets great output; another decides AI is useless. A shared prompt library fixes that. It captures the prompts that work, so every person starts from a proven pattern instead of reinventing it, and the firm gets consistent quality.
This guide is that library, organized by workflow. Use the prompts directly, adapt them to your firm, and treat the collection as a starting set you extend over time. The prompts assume an enterprise AI tool, which matters for the security reasons in section three.
2. The Rules of a Good PE Prompt
Every effective prompt below shares four ingredients. Learn the pattern and you can write your own.
Role. Tell the model who it is. "You are a PE deal screener for a lower-middle-market buyout fund" produces sharper output than no framing.
Context. Give it the situation and the constraints: the sector, the stage, what you care about, what to ignore.
Format. Say exactly what you want back: a five-bullet summary, a table, a 250-word section, a scored list. Vague asks get vague answers.
The document. Wherever possible, give the model the actual source (the CIM, the contract, the financials) rather than relying on its memory. Grounded prompts are more accurate and far less prone to invention.
A fifth, non-negotiable habit: verify. Every number and every factual claim gets checked at the source before it goes anywhere. The prompt gets you a fast, well-structured draft. You own the truth of it.
3. The Security Rule Before Any Prompt
One rule governs all of this: never paste confidential deal data into a consumer AI account. On free and personal tiers, your input can be used to train the model. A CIM, a target's financials, or LP data does not belong there.
Use a sanctioned enterprise tool (ChatGPT Team or Enterprise, Microsoft Copilot in your tenant, or Claude Team or Enterprise) that does not train on your data. The prompts in this library assume that environment. The capability is the same; the data handling is what protects the firm.
If you are unsure which tools clear the bar, our assistant comparison and Security and Data Governance guide cover it. Get the tool right first, then use the prompts freely.
4. Deal Sourcing and Screening Prompts
For evaluating inbound opportunities and preparing to engage.
"You are a PE deal screener for a lower-middle-market buyout fund focused on [sector]. Here is a CIM [attach]. Summarize the business in five bullets, then score it 1 to 5 against each criterion: recurring revenue, margin profile, customer concentration, fragmentation for buy-and-build, and management depth. List the three things I should diligence first and why."
"Here is our investment thesis for [company] [paste]. List the eight risks most likely to break this thesis, ranked by likelihood and impact, with one sentence each on how I would test it in diligence."
"Draft 15 questions for a first management call with a [sector] business doing roughly $[X]M EBITDA, focused on the durability of revenue, the biggest downside risks, and the owner's reason for selling. Group them by theme."
"Explain the [sector] industry to me as if I am about to meet a management team: the business model, how players make money, the main competitive dynamics, and the three trends shaping it. Flag anything I should verify with an expert."
5. Due Diligence Prompts
For reading the data room faster and sharper. Always confirm findings at the source.
"Read this agreement [attach]. Extract, with the exact language and the section reference for each: change-of-control provisions, assignment restrictions, termination rights, exclusivity, and any most-favored-nation clauses. Note anything that looks off-market."
"Summarize this quality of earnings report [attach] in one page for an IC audience. Highlight the adjustments that most affect normalized EBITDA, and flag any adjustment you would push back on, with your reasoning."
"Here is three years of customer revenue data [paste]. Calculate the top 1, 5, and 10 customers as a percentage of revenue for each year, flag any customer that grew or shrank more than 25 percent year over year, and note the concentration risk in two sentences."
"For a [sector] business we are acquiring, draft a commercial due diligence checklist organized by area (market, customers, competition, pricing, growth), with the specific questions and the data we should request for each."
6. Financial Analysis Prompts
For the model and the numbers. The model writes formulas and explains; you verify every figure.
"Write an Excel formula that calculates levered IRR from the cash flow series in cells B2 through B12, where B2 is the entry equity as a negative number. Explain what the formula does in one line."
"Describe the steps to build a two-way data table in Excel that flexes entry multiple (6x to 10x) against revenue CAGR (3 to 9 percent) and outputs five-year equity IRR. Tell me exactly which cells to set as the row and column inputs."
"Here is a list of proposed EBITDA add-backs [paste]. Categorize each as clearly defensible, arguable, or aggressive, with one sentence of reasoning. Flag any that an experienced buyer would challenge."
"Here is the company's P&L for the last three years [paste]. Walk through the biggest drivers of the change in gross margin year over year, and list the three questions I should ask management about it."
7. IC Memo and Writing Prompts
For drafting investment materials. The draft is a starting point; the deal team owns the argument.
"Draft the Business Overview section of an IC memo from these notes [paste]. Neutral, institutional tone, 250 to 300 words, no marketing language, no superlatives. Lead with what the company does and how it makes money."
"Turn these model outputs [paste base, upside, downside IRR and MOIC] into a Returns Summary paragraph for the IC. State each case, the key driver of each, and the single most important assumption the return depends on."
"Rewrite this risk section [paste] to be sharper and more specific. Remove hedging language, name the actual risk and its mitigant, keep it under 200 words, and do not soften the real concerns."
"You are a skeptical IC member. Read this memo [paste] and write the three hardest questions you would ask, and the strongest argument for passing on this deal."
8. Portfolio Monitoring and Value Creation Prompts
For operating partners and portfolio teams.
"Here are this month's financials for a portfolio company [paste]. Identify the three metrics trending worst versus plan, the likely drivers, and four questions for the next board call. Keep it to one page."
"Given this company's P&L [paste], list the five highest-probability margin improvement levers a PE operating partner would test in the first 100 days, ranked by likely EBITDA impact and ease of execution."
"Summarize these monthly reports from [N] portfolio companies [paste] into a one-page quarterly update for the investment team. Flag any company that breached or approached a covenant, and any that materially beat or missed plan."
"Here is the company's pricing and volume data by segment [paste]. Identify where the company appears to be underpricing, where discounts are deepest, and the two pricing changes most likely to add margin without losing material volume."
9. Investor Relations and Fundraising Prompts
For LP communication and fundraising. A human owns every word an LP reads.
"Draft a quarterly LP update from these portfolio highlights [paste]. House tone: direct, specific, no jargon, no hype. About 400 words. Lead with fund performance, then notable portfolio developments, then outlook."
"Here is an LP due-diligence questionnaire [attach] and our approved answer library [attach]. Draft an answer to each question using only the library. Flag any question the library does not cover so I can write it myself."
"Rewrite this fund overview copy [paste] to be specific and confident. Replace generic phrases with concrete numbers and facts, keep our tone, and remove anything that sounds like every other fund."
"We are meeting [LP type] who allocates to [strategy]. Based on this fund summary [paste], list the eight questions they are most likely to ask and a crisp, honest answer to each."
10. Turning Prompts Into Custom GPTs
A prompt you use every week should not be retyped every week. The next step up is to bake it into a custom GPT (on ChatGPT Team or Enterprise) or an equivalent assistant.
Take your best screening prompt and build a "Screening GPT" loaded with your investment criteria, so an analyst just uploads the CIM. Take your memo prompts and build a "Memo GPT" primed with your firm's structure and tone. Take your DDQ approach and build an assistant grounded in your approved answer library. The prompt becomes a tool the whole firm shares, with the firm's standards built in.
This is the bridge from a prompt library to something more durable. The principles are the same, and the security tier still matters; custom GPTs on Team and Enterprise do not train on your data. We cover building these in the ChatGPT for PE guide, and the step up to true custom agents in the AI for Excel guide.
11. Rolling Out a Firm Prompt Library
Turning this into a firm capability rather than a personal trick.
First. Make sure everyone is on a sanctioned enterprise tool, so confidential prompts are safe.
Second. Start a shared library: collect the prompts above that fit your workflows, and add the ones your best users discover. Keep it somewhere everyone can reach.
Third. Promote the most-used prompts into custom GPTs so the firm's standards are built in, and verify outputs as a matter of habit.
A Discovery Sprint can build a tailored prompt library and the custom GPTs around your firm's actual workflows, so the whole team gets consistent, safe, high-quality output instead of uneven ad-hoc results.
"The gap between people getting little from AI and people getting a lot is mostly skill in how they direct it. The technology is jagged: extraordinary on some tasks, weak on others, and knowing how to prompt and where to apply it is what separates the two groups."
Harvard Business School, field study on generative AI and knowledge work (2023)
- •Prompting is a learnable skill, and a shared library gives the whole firm consistent quality instead of uneven ad-hoc results.
- •Every good PE prompt has four parts: a role, the context, the format you want, and the source document itself.
- •Always verify. The prompt gives you a fast, structured draft; you own the truth of every number and claim.
- •The overriding rule: never paste confidential deal data into a consumer AI account. Use a sanctioned enterprise tool that does not train on your data.
- •The library spans the deal lifecycle: sourcing and screening, diligence, financial analysis, IC memos, portfolio monitoring, and IR.
- •Promote your most-used prompts into custom GPTs so the firm's criteria, structure, and tone are built in and shared.
- •Roll it out by getting everyone on a safe tool, building a shared library, and turning the winners into custom GPTs.
Related Guides & Articles
Want a prompt library built for your firm?
A Discovery Sprint builds a tailored prompt library and custom GPTs around your actual workflows, so the whole team gets consistent, safe, high-quality output. Start by finding your best use cases with the AI Use Case Finder.
Book a Discovery Sprint