GuidesAINovember 5, 2026Dennis Raisich

AI in due diligence: what it checks today

In due diligence, AI mainly takes over the reading. It sorts the documents, reads contracts and financial statements, cross-checks them against each other and writes a first draft of the report. The judgment and the approval stay with the advisors.

In short

AI recognizes and sorts documents, extracts details from contracts, recalculates key figures and finds contradictions between documents.
An AI statement is only useful if it is backed by evidence: every statement must lead to its source in the documents.
When choosing a platform, what counts is evidence, data location and no training on your documents, plus roles and logging.
Responsibility for the report stays with humans.

What does AI take over in due diligence?

Sorting documents

Data rooms rarely arrive sorted: PDF, scans, Excel, Word and emails, often as whole folders or a ZIP file. AI recognizes what a document is, for example a lease, annual accounts or an employment contract, and files it in the right area. Text recognition makes scans readable.

Finding missing documents

Which documents a review needs can be derived from the country, legal form and industry. When AI checks this list against the data room, you can see at once what is missing and needs to be requested from the seller.

Reading contracts

AI pulls the key details from every contract into a table: parties, term, termination, change of control, exclusivity, liability. This approach is called tabular review. Instead of opening a hundred contracts one by one, you ask all of them one question, for example: “Which contracts can the other party terminate if the company changes owners?”

Recalculating the numbers

AI reads revenue, EBITDA and net debt directly from the financial statements, over several years and for every company in a group, and recalculates them. Unusual items are recorded as findings. More on this in our guide to financial due diligence.

Reading documents against each other

This is where the biggest gain lies. A human usually reads a contract and a business plan on different days, and often two different people read them. AI puts them side by side. A fictional example: 42% of revenue depends on one customer whose framework agreement can be terminated on a change of ownership, yet the business plan carries this revenue forward in full. Contradictions like these between contracts, financial statements and the plan are the findings that change a purchase price.

Drafting the report

Finally, AI writes a draft of the due diligence report: management summary, findings by area, recommendations. The advisors review, edit and approve it.

Where does AI reach its limits?

Language models can make up statements that sound plausible. In due diligence, that is unacceptable, because the purchase price and the warranties rest on the report. That is why one simple rule applies: a statement without a source in the documents does not belong in the report.

AI also only knows what is in the data room. Talks with management, site visits and knowledge of the market and the industry remain part of the review. Whether a risk breaks the deal or can be cushioned in the purchase agreement is for the parties involved to decide. Machines, too, read blurry scans and handwriting less reliably, and a gap in the data room remains a gap.

What should you look for in an AI platform?

Evidence. Does every statement lead to its source in the document, and can you read it there?
Data location. Where are the database and files stored, and where does the AI run? For lawyers and accountants, professional secrecy also applies.
Training. Does the contract rule out your documents being used to train AI models?
Permissions and logging. Does each role only see what it is allowed to see, and is every access logged?
Coverage. Does the platform review legal, tax and financial matters, or only contracts?
Collaboration. Do sellers, bidders and specialists work in the same data room?
Result. Do you get a report that you can review, edit and share?

How does majiq do due diligence with AI?

majiq is an AI due diligence platform from Switzerland. majiq reads every document in the data room, checks it in context and writes the due diligence report. Every finding is backed by evidence from your documents and cross-checked by two AI reviewers. You accept or reject the AI’s suggested changes.

The database and files are stored in Switzerland, with hosting in Zürich. Text recognition and AI analysis run in the EU, and your documents don’t train any AI models, as set out in the contract. Results are ready in hours, not weeks; how long it takes in each case depends on the size of the data room. The sample report shows what a report looks like, and all features are listed under Features.

Frequently asked questions

Does AI replace lawyers or accountants?

No. AI takes over the reading and writes a draft. The judgment, the negotiation and the responsibility for the report stay with the advisors.

Am I allowed to upload a client’s documents to an AI?

That depends on the engagement agreement, professional rules and data protection law. Check where the provider stores and processes the data, whether it uses the data for training and whether a data processing agreement is in place.

How reliable is AI in due diligence?

As reliable as its evidence. An AI statement only becomes verifiable when it leads to its source and you can read the sentence in the original.

What is a tabular review?

A tabular review moves many documents of one type into a table: one row per document, one column per detail, such as the term or the notice period. That way you can compare a hundred contracts without reading each one individually.