Financial and transaction analysis
Ledgers and financial records classified against the terms that matter, with the other side's math reproduced so it can be tested line by line.
Case support your attorneys can check.
When a matter turns on thousands of records, I do the analysis with AI alongside your team: classifying transactions, reproducing the other side's math, turning email archives into review sets. Every result traces back to its source.
What I take on
The work a matter turns on.
Analysis too large to do well by hand, done with AI and checked before anyone relies on it.
Email and document collections
Outlook and MBOX archives turned into individual documents, with attachments extracted and an index your team can review, every entry linked to its source.
Chronologies and issue maps
Timelines and issue structures built from thousands of emails, records, and filings, each entry traceable back to its source.
Discovery, exhibits, and transcripts
Productions processed before the deadline, exhibits organized for trial, and deposition transcripts summarized across a multi-party case.
When firms call me
A matter that can't wait.
- A financial record the case turns on, too large to review well by hand
- An email archive that has to become a review set
- A discovery production that has to be processed before a deadline
- A timeline that has to be built from thousands of emails or records
- Exhibits to organize for trial, or transcripts to summarize across a multi-party case
- Due diligence review on a compressed deal timeline
How the work runs
Scope, do the work, check it, hand it over.
The checking step is the one most AI work skips, and it decides whether the result can be trusted.
01
Scope the question and the boundaries
I pin down with your team what the analysis has to answer, what good output looks like, and where the data may and may not go. The boundaries are written down before any work starts.
02
Do the work on your records
I run the analysis with AI on your records and keep every result tied to the document it came from, so nothing has to be taken on faith.
03
Check it before anyone relies on it
Where the stakes call for it, a second, independent pass rebuilds the analysis blind. The method is written down, so the result can be reproduced.
04
Hand over work your attorneys can check
You get the results, the method, and any tool I built. I'm not an attorney: your lawyers decide what the work supports.
Results
What it produced for two firms.
Recent engagements, described without client names.
Case support · Insurance premium dispute
A seven-figure premium audit, rebuilt from the ledger up.
For a litigation team disputing the audit, I classified more than 280,000 general-ledger transactions against the policy's exclusions, reproduced the insurer's own premium math to the dollar, then had a second AI system rebuild the analysis blind to check the first.
85 to 90% below the premium the insurer claimed, in a fully reproducible, audit-ready record
Case support · Email archive review
Email archives, turned into a review set a paralegal can run.
I built the firm a tool that runs entirely on its own computer. It converts Outlook and MBOX archives into individual email documents, extracts every attachment, and builds an Excel index with a summary of each message and a link to its file.
Minutes to process tens of thousands of emails, with client data kept on the firm's own machine
Your data, your judgment
Where the data goes, and who decides.
Client material stays inside tools your firm has approved, under your accounts, or on your own computer, as with the email tool above. I do the analysis. Your lawyers decide what it supports.
Let's talk
Have a matter that turns on a large record?
Tell me the matter, the record volume, and the deadline. I'll tell you quickly whether and how I can help.
Prefer to reach me directly? steve@intelligencebyintent.com (805) 876-4847
Please don't include confidential client information in an email.