One missed risk
can change
the deal.
All3 is building AI for legal due diligence—from the initial request list to the final report, guided by a library of expected Brazilian legal risks.
Before you buy or invest in a company, someone must read its documents and explain the risks that come with the deal. Today, that work depends on a fragmented, labor-intensive process.
The process is expensive.
The missed risk can cost far more.
Law firms are one missed risk away from irreversible financial or reputational damage.
A junior lawyer may coordinate several deals at once, reviewing at 2 a.m. for a 6 a.m. deadline. The risk is an overlooked issue that changes what the buyer actually owns.
Large-engagement scenario described in the founder’s one-pager.
The protected-land example in the one-pager
Founder-reported; not independently verified. The one-pager describes a $1B acquisition in which $400M was tied to land that could not be developed. It says the buyer discovered the restriction after closing and sued its counsel. On those figures, the remaining value is $600M. No firm or transaction is identified here.
100 hours can disappear into formatting and consolidating reports.
Illustrative economics and hours from the one-pager; not a measured market average or an All3 savings claim.
The hard part lives
inside private deals.
Legal diligence requires practical deal knowledge, coordination and judgment. Document analysis addresses only part of the work.
The knowledge is private
Reports, Q&A and deal documents are confidential. Lawyers learn what matters through years of transactions. That practical knowledge is difficult to assemble into a public dataset.
Plausible risks can be the wrong risks
A general model can produce convincing but irrelevant flags while missing material issues. Across thousands of documents, the noise competes with what lawyers need to see.
Hundreds of requests. Lawyers and company representatives going back and forth. A report that still needs experienced judgment.
This is the founder’s diagnosis of the problem. Whether a specialized risk library improves outcomes remains to be tested.
From the first request
to a report a lawyer can review.
The output separates the issue, legal risk and recommendation, with sources available for inspection. Lawyers remain responsible for reviewing the finding.
Actual screen · Click to enlargeRequests and responses sit in one workspace. The intended workflow connects this evidence to draft findings and the consolidated report.
Actual screen · Click to enlargeProduct screens show the interface. End-to-end execution and accuracy have not yet been validated with customers.
Tell AI what to look for
in each document.
All3’s approach starts with risks learned in Brazilian transactions. The library makes the review task specific.
“Find the legal risks
in this document.”
A broad search can surface risks that are plausible but immaterial to the deal.
Check the document against the risks expected in Brazilian diligence.
The one-pager’s example: focus a copyright review on the 15 risks the founder expects to arise in Brazilian diligence, instead of everything ever written about copyright.
Focused checks could reduce material omissions and irrelevant flags. If a general model does just as well, the library has not demonstrated an advantage.
The 15-check example illustrates the method; it is not a published benchmark or a claim of complete coverage.
Seven years mapping risks
and rebuilding the process.
Stanford GSB MBA ’28
University of São Paulo Law
“I started documenting how to fix diligence at 21.”
Victor mapped Brazilian diligence risks, rebuilt the process at Pinheiro Neto, wrote its playbook and trained hundreds of lawyers. All3 grows out of that work.
Seven years refers to mapping risks and improving the diligence process, not seven years of software development.
Start in Brazil.
Test who buys and why.
Legal-service spending, not potential software revenue. The estimates need a dated methodology.
The first customer is still a question
Large and mid-sized law firms. PE/VC funds. Corporate buyers. Pilots should establish who adopts and who owns the budget.
Pricing hypothesis
Charge per diligence engagement, including deals that do not close. Test the payer, the price and the value they receive.
The expansion thesis: the diligence workflow is similar across countries. The legal risks change. Expansion requires building and validating a risk library for each jurisdiction.
Why faster review does not automatically mean lower legal bills
Under a fee cap, fewer lawyer hours may improve a firm’s economics without reducing the buyer’s invoice. Pilots should measure lawyer effort, actual fees and the value each buyer receives separately.
Put the library
to the test on real deals.
No customers yet.
Six unpaid undergraduate volunteers have been recruited for market research.
A classmate will help with first sales, also unpaid.
Traditional diligence
A general-purpose model
Same real-deal scope. Compare accuracy, speed and cost. The library’s advantage must be earned in the results.
Six questions before go-to-market
Who adopts first?
Large law firms, mid-sized firms, or buyers such as PE/VC funds and corporates?
Which value drives adoption?
Speed, reliability, cost, or freeing lawyers from manual work?
Can PE funds drive adoption?
Would funds encourage or require their deal counsel to use All3?
Can bankers become a referral channel?
Could investment bankers introduce All3 to the buyers and lawyers on their transactions?
What pricing model works?
Who pays, how much, and on what basis? Per-engagement pricing is one hypothesis.
How much human review builds trust?
What do lawyers need to inspect and approve before customers will rely on the output?
I’d value your feedback on what to test first—and what a convincing result would look like.Stanford feedback discussion · Victor Pires · vpires@stanford.edu