Bramwell

Reads a fund agreement in minutes and shows where every answer came from.

Year: 2026
Status: Delivered to a client

Case study on Slipstream Systems

The problem

The terms that decide whether a deal is any good are buried across dozens of pages.

Reviewing a private equity fund agreement means hunting through the whole document for the fees, carry, preferred return and investor protections, then judging each one against what the market usually allows. It is slow, skilled work, and it is easy to miss the one clause that matters. The people who commissioned this wanted to upload an agreement and get back a rated list of the terms that matter, with the red flags called out.

The Bramwell scorecard for a sample fund agreement, showing a color bar of rating counts, the most severe findings ranked at the top, and each row giving the term, the document's language, the market benchmark and the rating. — The scorecard. Every key term rated, the counts and color bar summarizing the agreement at a glance, and the worst findings first. Screenshots use a fictional sample agreement, "Crestline Partners Fund III"; no client documents are shown.

What it makes possible

01
A rated scorecard instead of a reading assignment
Every key term gets one of four ratings: red flag, unfavorable, standard or favorable. The counts and color bar show the overall picture at a glance and the most serious issues rise to the top. On the sample agreement that surfaced 18 red flags, including a GP who could amend the agreement without investor consent.
02
Every rating points at its clause
Click a page number and the original agreement opens beside the scorecard with the exact sentence highlighted. Each citation is checked against the real text before it is shown, so a reviewer can verify any finding in seconds rather than take it on faith.
03
Organized the way reviewers already think
Findings are grouped into 11 sections, from fund basics and management fees through carried interest, expenses, governance and NAV facilities. Each row shows what the document says, the market benchmark, the rating and the reason, and the whole thing can be filtered down to what differs from standard.
04
Ask the document a question
Reviewers can question the agreement in plain English, and Bramwell suggests starter questions built from its own findings, including points to negotiate with the GP. Answers cite the clauses they rely on, keep the document's own language separate from general market context, and calculate fee and waterfall math directly rather than letting the model estimate it.
05
Sensitive files never leave the machine
It installs as a one-click Windows desktop app. Documents and results are stored locally and encrypted, and document text goes only to the AI provider for analysis under the client's own account, which the app says up front. No terminal, no server, nothing of theirs sitting on mine.
06
A report you can hand to a partner
One click exports the full scorecard as a PDF, an Excel workbook with live formulas, Markdown or JSON, with every page reference marked as verified.

54Deal terms checked against market benchmarks
$0.84AI cost to analyze the sample agreement
18Red flags found in the sample agreement

In use

A flagged fund-term finding with the original agreement open beside it and the sentence in Section 2.4 highlighted that lets the GP extend the term by three years on its own. — Every rating shows its source. Clicking the page number opens the agreement with the exact sentence highlighted; here the fund term is flagged because the GP can extend it by three years alone, and Section 2.4 is where that comes from.

The findings table grouped into sections such as management fees, carried interest and governance, each row showing the document's language, the market benchmark, the rating and the reason. — Findings grouped into the 11 sections a reviewer would already be thinking in, with filters for rating, search and "only what differs from standard."

A plain-English question about the agreement answered with cited clauses, each citation marked as verified against the document text. — Answers that cite the document. Each citation is checked against the actual text and opens it highlighted.

The export dialog offering the scorecard as PDF, Excel, Markdown or JSON, with a note that every page reference has been verified. — Reports ready to share: PDF, Excel with live formulas, Markdown or JSON.

Notes

Where it came from

This one was commissioned. A group reviewing fund documents and single-company SPVs described what they wanted in one sentence: upload an investment agreement, get back a list of predetermined criteria rated fair, unfair or standard, and have the red flags highlighted. Bramwell is that sentence built out, delivered and installed on their machines, and the first piece of work done under Slipstream Systems, the consulting business I started around it.

The part worth pointing at

The hard requirement was not the analysis. It was trust. A reviewer who has to re-read the whole agreement to check the tool’s work has gained nothing, so every rating had to point at the clause it came from, and every citation had to be verified against the real text before it was shown. The same principle runs through the Q&A: the document’s own language is kept apart from general market context, and anything with numbers in it, fees, waterfalls, is calculated rather than estimated.

The second requirement was confidentiality. These are documents nobody wants on a stranger’s server, so there is no server. It runs on the client’s own computer, stores everything locally and encrypted, and is explicit about the one place document text does go.

Where it stands

Delivered and in use. The next phase, if the client wants it, is comparing several funds side by side and extending the Q&A across a whole portfolio of agreements rather than one at a time.