Due Diligence Engine
AI-powered technical due diligence for venture capital. Verify a startup's technical claims, detect AI-washing, and score technical risk with confidence — straight from your IDE.
Technical due diligence, run by the AI you already have
The Due Diligence Engine (DDE) is an open-source command-line tool that turns the AI agent in your IDE — Claude Code, Cursor, GitHub Copilot and others — into a world-class technical due-diligence analyst. It reads a startup's entire codebase, researches the market live, evaluates the project across multiple weighted dimensions, and writes a consulting-grade PDF report to your Downloads folder. Everything runs locally inside your IDE's AI sandbox: no API keys, no third-party servers, no telemetry, and no added cost.
In the AI era, claims outpace reality
Every deck says "AI-powered." Few can prove it. Technical diligence is slow, expensive, and hard to scale across a pipeline.
AI-washing is everywhere
Marketing language exaggerates — or invents — capabilities the codebase does not support. Spotting the gap takes a senior engineer's time.
Manual review doesn't scale
Reading a codebase, mapping the architecture, and benchmarking competitors for every deal is days of expert work you rarely have.
Hype distorts valuations
Without an objective technical baseline, pricing and risk decisions lean on the founder's narrative instead of the evidence.
From one install to a 24-page report
DDE generates a structured prompt that any file-reading, shell-running AI agent can execute. No DDE-specific plugin or login required.
Install from the public repository with a single pip command. Pure local Python — no accounts, no configuration.
From the target project, run the DDE command in your AI IDE terminal. Your existing AI subscription does the analysis — at no extra cost.
The agent reads hundreds of files, evaluates across weighted dimensions, and researches competitors, funding and security advisories live on the web.
A consulting-style PDF lands in your Downloads folder — a 0–100 score, a letter grade, a competitive landscape, and sourced rationales.
What DDE evaluates
A repeatable, evidence-first technical assessment — the same rubric across every deal.
Claim verification
Cross-checks the pitch against the actual codebase to confirm what is really built versus what is merely asserted.
AI-washing detection
Flags where "AI" is a thin wrapper, a hardcoded heuristic, or absent entirely — separating genuine ML from marketing.
Confidence-scored grading
Produces a 0–100 score and a letter grade across equally weighted dimensions, with the reasoning shown.
Codebase & architecture review
Maps structure, dependencies and design choices to assess maintainability, scalability and technical debt.
Live competitive landscape
Researches real competitors, funding and security advisories on the web and builds a sourced market comparison.
Local & private by design
Runs entirely inside your IDE's AI sandbox. No third-party servers, no telemetry — audited with CodeQL, Dependabot and pip-audit.
Built for people who write checks on technology
Anyone who needs an objective read on a startup's engineering — fast.
VC & growth funds
Screen technical risk across a pipeline before committing partner hours to deep diligence.
Angels & accelerators
Get a senior-engineer-grade second opinion without a senior engineer on staff.
Corp dev & M&A
Baseline an acquisition target's codebase and architecture before the technical deep-dive.
Frequently asked questions
Is Due Diligence Engine free?
Yes. DDE is open source under the Apache 2.0 license. It uses the AI subscription you already have in your IDE, so there is no API key and no added cost.
Does my code leave my machine?
No. DDE is 100% local Python and runs inside your IDE's AI sandbox. There are no DDE servers and no telemetry. Web research only fetches public information about competitors and the market.
Which AI tools can run DDE?
Any agent that can read files and run shell commands. It is verified with Claude Code and GitHub Copilot Agent Mode, and works with Cursor, Gemini Code Assist, Aider, Windsurf and similar agentic terminals.
What does the output look like?
A consulting-grade PDF (around 24 pages) saved to your Downloads folder, with an overall 0–100 score, a letter grade, a competitive landscape, an implementation matrix, and sourced rationales.
What is "AI-washing" and how does DDE detect it?
AI-washing is marketing a product as AI-powered when the underlying technology is minimal or absent. DDE compares the claims in the pitch against the real implementation in the codebase and flags the gaps.
Run your first report today
Install once, run from any AI IDE, and read the evidence for yourself. Free and open source.