What it is
Source: My own description of the project; the Corporate Finance Institute for the definition of due diligence. Date: 15 October 2026. Scope: What the system does, which of its agents work, and how they find and rate what they report. No accuracy, speed or deal outcome is claimed, and no client or deal is named.
The Enterprise AI Due Diligence Agent is a multi-agent system, written in Python, that turns deal documents into evidence-backed reports. Built
Due diligence is the process of verifying, investigating or auditing a potential deal or investment to confirm the relevant facts and financial information. It is completed before the deal closes (Corporate Finance Institute).
What it does
It has four working agents. Built
- Extraction. Pulls the information out of the deal documents.
- Financial analysis. Analyses the financial information in them.
- Cross-document contradiction detection. Checks the documents against one another for places where two of them disagree about the same matter.
- Risk engine. Rates risk with a transparent likelihood × impact model.
Every finding in a report cites its source.
How it is built
It is written in Python as a multi-agent system, which means the work is divided between several agents, each with its own task. Contradiction detection is deterministic, with no language model in its path, so the same documents always give the same contradiction findings. Built
Stack. Python, multi-agent
Where it stands
Four agents work so far. Built No accuracy figure, speed or deal outcome is claimed for it.
My part
I designed the system and built its four working agents in Python. Built
