Most investment firms experimenting with AI make the same mistake.
Every few weeks someone posts a thread about building AI agents for private equity, lists the ten agents they built, racks up four hundred comments, and then makes you sign up for a workshop to get the prompts. The idea they are sharing is genuinely good. The bit where you have to trade your email to see the prompt is the part we find a little annoying.
So we are just going to give you the prompts. All of them. We build this software for a living, which means we have written these prompts more times than is probably healthy, and we would rather you see how the prompt is made than guard it like a trade secret.
This guide walks through the ten AI agents increasingly becoming part of modern due diligence workflows. Every section includes a practical prompt you can use today in ChatGPT, Claude, Gemini or another capable model.
Why Multiple AI Agents Beat One General AI Assistant
The biggest shift isn't better prompting. It's better workflow design. Ask a single model to "run due diligence on this company" and you get something that sounds like diligence and contains almost no diligence. Confident, fluent, useless. The reason is simple once you see it: diligence is not one task.
Investment diligence is multidisciplinary by nature. Asking one AI assistant to perform every task at once forces it to balance competing objectives, maintain too much context and switch constantly between financial analysis, commercial reasoning, legal review and operational assessment.
The result is often a polished report with surprisingly little investment insight.
Breaking the work into specialised AI agents solves that problem.
Each agent has:
- one clearly defined objective
- one output format
- one evidence standard
- one area of responsibility
Together they produce a diligence process that's easier to verify, easier to audit and far easier to challenge before an investment committee meeting.
Two principles matter more than anything else.
1. Every agent should work from the same evidence base
Commercial, financial and legal analyses should reference the same data room, management materials, financial statements and customer information.
Without a shared source of truth, agents quietly contradict one another. Those inconsistencies often surface only after weeks of work or worse, after closing.
2. Every claim should be traceable
AI shouldn't be rewarded for sounding confident.
It should be required to cite evidence.
Every prompt in this guide instructs the model to reference specific documents, identify missing information and clearly distinguish verified facts from assumptions. That single change dramatically reduces hallucinations and makes the outputs much more useful for professional investors.
The prompts below assume you've already provided the relevant diligence materials as context. Replace the placeholders with your own documents and adjust the output format where appropriate.
1. Commercial Due Diligence AI Agent
Commercial diligence answers the question every investment eventually comes back to:
Why does this business win?
Revenue growth alone rarely tells the full story. Strong businesses grow because they have durable competitive advantages, defensible markets and customers who are difficult to displace. Weak businesses often grow for reasons that disappear after acquisition.
A commercial AI agent evaluates those fundamentals while highlighting where the investment narrative extends beyond the available evidence.
2. Financial Due Diligence AI Agent
Financial diligence is where good stories meet bad arithmetic.
Management presentations are designed to explain performance. Financial diligence is designed to verify it.
A dedicated financial agent reconstructs the numbers from source documents instead of accepting headline metrics at face value. It looks for inconsistencies across financial statements, quality-of-earnings issues, aggressive revenue recognition and EBITDA adjustments that deserve a second look.
The goal isn't to predict the future. It's to establish whether the past is being represented accurately.
You are a financial due diligence analyst. From the attached financials, reconstruct revenue, gross margin, and EBITDA for each period. Flag: revenue recognition that looks aggressive, one-off items added back to EBITDA, margin movements you cannot explain from the documents, and any figure that does not reconcile across two documents. Do not estimate; if a number is not in the materials, mark it UNVERIFIED. Output: reconstructed summary table, list of red flags ranked by materiality, list of UNVERIFIED items.
3. CIM Analysis AI Agent
Every Confidential Information Memorandum is written to sell an investment opportunity.
That's its job.
The challenge for investors is separating evidence from narrative.
A CIM analysis agent systematically breaks apart the seller's investment thesis, verifies each supporting claim against the underlying documents and highlights where the story relies more on positioning than proof.
This isn't about catching anyone out. It's about identifying the assumptions you're implicitly underwriting.
You are reviewing a Confidential Information Memorandum (CIM). Extract the seller's core investment thesis in three bullets. Then, for each claim, determine whether the attached data SUPPORTS, PARTIALLY SUPPORTS, or DOES NOT SUPPORT it, citing the relevant document and page. Identify the three places where the narrative appears to compensate for weak evidence. Output: investment thesis summary, claim verification table, narrative risk assessment.
4. Management Due Diligence AI Agent
Most investments succeed, or fail, because of people.
Financial models don't execute strategy. Leadership teams do.
Management diligence is notoriously difficult because the evidence is qualitative. Interviews, organisational structures, hiring patterns and historical decisions often reveal more than polished biographies.
An AI agent can't judge character.
It can, however, surface patterns that deserve attention before you walk into a management meeting.
You are assessing the management team using the attached executive biographies, organisation chart and interview notes. Identify key-person dependencies, leadership gaps relative to the company's current stage, inconsistencies between management statements and operational evidence, succession risks and indicators of organisational turnover. Cite supporting evidence throughout. Output: leadership strengths, ranked management risks and the three highest-priority questions for the next management session.
5. Customer & Market Intelligence AI Agent
Revenue tells you what customers did.
Customer diligence tells you why.
Businesses rarely lose customers overnight. The warning signs usually appear months earlier through shorter contract terms, increasing concessions, weakening engagement or slowing expansion revenue.
A market intelligence agent pieces together those signals across customer interviews, churn reports, CRM exports and market research to assess whether demand is durable or simply benefiting from favourable conditions.
It's often the difference between buying a category leader and buying a temporary winner.
You are a market intelligence analyst. Using the attached customer references, churn reports, CRM exports and market research, assess the durability of customer demand. Identify churn patterns masked by new-logo growth, signs of weak customer commitment, pricing pressure and whether the stated addressable market is supported by the available evidence. Cite all findings. Output: demand durability assessment, supporting evidence and the single biggest unresolved market risk.
6. Valuation & Scenario Analysis AI Agent
Valuation models are only as reliable as the assumptions underneath them.
AI can calculate scenarios almost instantly.
It can also invent market multiples with alarming confidence if you let it.
A good valuation agent behaves more like a disciplined associate than an eager analyst. Every assumption is explicit. Every input is documented. Every sensitivity is visible.
If the evidence doesn't exist, the model says so instead of filling in the gaps.
That's exactly what you want.
You are a valuation analyst. Using the attached financial statements and assumptions, prepare base, downside and upside operating scenarios covering the next three years. Document every assumption and reference supporting evidence wherever possible. Identify the two assumptions with the greatest impact on valuation. Do not invent market multiples or comparable company data; if they are unavailable, flag them as missing. Output: three-scenario valuation summary, sensitivity analysis and assumptions log.
7. Portfolio Monitoring AI Agent
Due diligence doesn't end when the deal closes.
In many ways, that's when it becomes more valuable.
Portfolio companies generate a constant stream of board packs, financial reports, KPI dashboards, customer updates and operating metrics. The challenge isn't collecting that information—it's spotting what changed before it becomes a board-level problem.
A portfolio monitoring agent compares each reporting cycle against historical performance, budget and investment assumptions, surfacing deviations that deserve attention.
Instead of reading every board deck from cover to cover, investment teams start with the handful of changes that actually matter.
You are monitoring a portfolio company. Using the attached board packs, financial reports and operating metrics, compare actual performance against both the previous reporting period and the approved operating plan. Flag covenant proximity, cash runway changes, customer concentration shifts, leadership changes and any KPI moving in the wrong direction by more than [X]%. Output: a traffic-light dashboard (Green, Amber, Red), the two highest-priority issues this quarter and the additional information required before the next board meeting.
8. Investment Committee Memo AI Agent
The investment committee memo isn't another diligence document.
It's the document that forces every workstream into a single investment recommendation.
By this stage, commercial diligence, financial review, legal analysis and operational assessment should already exist. The IC memo shouldn't repeat them. It should synthesise them.
That distinction matters.
Good investment memos don't contain more information. They contain better judgement.
An AI agent can accelerate the drafting process—but only if it's restricted to verified findings rather than generating new conclusions.
You are preparing an Investment Committee memo. Using only the verified outputs from the attached diligence workstreams, produce: (1) an investment recommendation, (2) the investment thesis in three points, (3) the three most material investment risks and proposed mitigants, (4) a valuation summary, and (5) a clearly labelled list of open diligence items that must be resolved before signing. Do not introduce any new claims that are not supported by the attached analyses. Keep the memo under two pages.
9. Legal & Compliance AI Agent
AI won't replace legal counsel.
It can make legal review dramatically more efficient.
Most transactions involve hundreds of agreements spread across customer contracts, employment documents, shareholder agreements, licensing arrangements and regulatory filings. The first job isn't interpreting every clause, it's finding the ones worth escalating.
A legal diligence agent acts as an intelligent first pass, surfacing contractual obligations and missing documentation long before external counsel starts billing by the hour.
You are a legal due diligence reviewer. Using the attached contracts, corporate records and regulatory documents, identify change-of-control clauses, exclusivity provisions, non-compete obligations, unsigned agreements, pending litigation, missing corporate records and regulatory licences that may affect the transaction. Cite the relevant document and assess the severity of each issue. Output: prioritised legal issues, supporting citations and a list of additional documents required.
10. Operations Due Diligence AI Agent
Strong businesses don't just sell well.
They deliver consistently.
Operations diligence examines whether the business can support future growth without depending on heroic effort, fragile systems or a handful of critical people.
It's also where many integration risks become visible before close.
Supplier concentration, technology debt, cybersecurity posture, fulfilment bottlenecks and operational resilience rarely dominate the CIM, but they often determine how much value an acquirer can actually realise after the transaction.
You are an operations due diligence analyst. Using the attached operational documentation, assess operational scalability, supplier dependencies, technology architecture, cybersecurity posture, business continuity and potential single points of failure. Support every conclusion with evidence from the provided materials. Output: operational strengths, the three most material operational risks and post-acquisition integration considerations.
The Step Almost Every Team Skips
Running ten AI agents isn't the same thing as running AI-powered due diligence.
It just gives you ten independent analyses.
The real value appears when those analyses begin challenging one another.
Commercial diligence may conclude the company has exceptional pricing power.
The financial agent may discover margins deteriorating.
The customer agent may find increasing discounting hidden behind new-logo growth.
Each conclusion looks reasonable in isolation.
Taken together, they tell a very different story.
The best investment questions usually emerge from these contradictions—not from any single workstream.
That's why every multi-agent workflow should include one final synthesis step.
You are the diligence lead. Compare the outputs from the commercial, financial and management due diligence agents. Identify conclusions that contradict one another, assumptions that appear inconsistent and evidence gaps that materially affect the investment recommendation. Rank the contradictions by potential impact on the deal and recommend the additional diligence required before the investment committee meeting.
If you only adopt one idea from this article, make it this one.
Individual AI agents produce analysis.
Cross-agent review produces investment insight.
We've used variations of them ourselves, and they'll immediately improve the quality of AI-assisted diligence.
Eventually, though, most investment teams run into the same three bottlenecks.
1. You become the workflow
Every deal starts with copying documents into multiple chats, managing prompt versions, collecting outputs and assembling everything into a single report.
None of that creates investment insight.
It's administrative work disguised as analysis.
2. Your "shared evidence base" isn't actually shared
Ten browser tabs don't constitute a single source of truth.
Someone inevitably uploads a newer financial model into one conversation but not another. One agent references Version 3 while another analyses Version 2.
By the time the investment committee meets, nobody is completely certain which answer came from which document.
3. The system never learns
Every deal begins from scratch.
The AI doesn't remember your investment mandate.
It doesn't know your preferred diligence framework.
It doesn't recognise that your fund consistently prioritises capital efficiency over revenue growth, or net revenue retention over customer acquisition.
Institutional knowledge stays trapped inside the people running the process.
That's where prompts stop scaling.
From Prompt Library to Investment Infrastructure
This progression is why many firms eventually move beyond standalone prompts.
The objective isn't simply to automate individual diligence tasks.
It's to create an investment system where every workstream references the same evidence, every conclusion is traceable, and every new deal benefits from what the firm has already learned.
That's the philosophy behind askRIA.
Rather than running isolated prompts across multiple chat windows, askRIA coordinates specialised AI agents on a shared evidence layer that understands your fund's investment thesis, historical decisions, sectors, portfolio companies and diligence framework.
Commercial analysis, financial review, legal diligence, IC memo generation and founder evaluation all reference the same underlying information instead of operating independently.
The result isn't just faster due diligence.
It's more consistent decision-making.
The prompts in this guide are the manual version.
They're a practical place to start.
When your team reaches the point where managing the workflow takes more effort than the analysis itself, that's usually the signal that it's time for dedicated investment infrastructure.
Final Thoughts
AI isn't replacing due diligence.
It's changing where investors spend their time.
The repetitive work, reading hundreds of pages, extracting facts, reconciling numbers and drafting first-pass analyses, is increasingly becoming machine work.
Judgement isn't.
The best investment teams won't be the ones with the most AI agents.
They'll be the ones that ask better questions because the routine work has already been done.
That's a much more durable advantage than simply processing documents faster.
Keep reading
Run your first deal free in askRIA and get an IC memo in minutes.*
FAQs
- What are AI agents for due diligence?
AI agents for due diligence are specialised AI workflows that each perform a single diligence function, such as commercial analysis, financial review, legal diligence or investment committee memo preparation. Instead of relying on one general-purpose AI assistant, investment teams use multiple agents that work from the same evidence base to produce more accurate, verifiable analysis.
2. Why use multiple AI agents instead of one AI assistant?
Due diligence involves different disciplines that require different types of analysis. A commercial diligence agent evaluates market positioning, while a financial agent reconciles financial statements and a legal agent reviews contracts. Specialised agents produce more reliable outputs and can cross-check each other's findings before an investment decision is made.
3. Can I use these prompts in ChatGPT, Claude or Gemini?
Yes. The prompts in this guide are model-agnostic and can be used in ChatGPT, Claude, Gemini and other capable AI models. The main limitation is that you'll need to manage the workflow manually by providing documents, running each prompt separately and combining the outputs yourself.
4. What documents should I provide to an AI due diligence agent?
The quality of the output depends on the quality of the evidence. Typical inputs include:
- Confidential Information Memorandums (CIMs)
- Financial statements
- Data room documents
- Customer and market research
- Management interview notes
- Board materials
- Contracts and legal documents
- Operational reports
Providing a complete evidence base helps every AI agent produce more accurate and consistent results.
5. Can AI replace human due diligence?
No. AI accelerates document review, summarisation and structured analysis, but investment decisions still require human judgement. The best investment teams use AI to reduce repetitive work so they can spend more time validating assumptions, investigating risks and making better decisions.
6. How do private equity firms use AI in due diligence today?
Many firms use AI to analyse financials, review CIMs, identify legal risks, monitor portfolio companies, generate investment committee memos and surface inconsistencies across diligence workstreams. Rather than replacing existing processes, AI is increasingly being integrated into them to improve speed and consistency.
7. What is the biggest limitation of using AI prompts for due diligence?
Running individual prompts across multiple AI chats doesn't create a shared source of truth. Analysts still need to upload documents repeatedly, reconcile conflicting outputs and combine findings manually. As deal volume increases, workflow management often becomes the biggest bottleneck.
8. How does askRIA differ from using AI prompts manually?
Manual prompts help automate individual diligence tasks, but they don't retain institutional knowledge or coordinate work across multiple analyses. askRIA runs specialised AI agents on a shared evidence base that understands your fund's investment thesis, portfolio and diligence framework, enabling more consistent and scalable investment workflows.