Chorusse's practice work is advising firms on AI and client data. This is the other half: tools we build in the open, free, with the source readable by anyone. It exists partly because it's the work we like, and partly because it's the only honest way to show what we understand about these systems.
If you're deciding whether to let someone near your firm's client-data questions, credentials only tell you so much. What's harder to fake is a body of work that's public, specific, and candid about where it falls short — including a tool of ours that measures itself against a competitor and reports when it added nothing.
A panel of expert lenses pressure-tests a hard call and hands you a one-page brief — the ranked recommendation, the confidence, and the dissent.
A decision doesn't end when you make it. Memory keeps the brief and brings it back when the same call comes round again.
Including a section on where the competitor is genuinely better than us. The empirical half is pre-registered and hasn't been run yet — and rather than invent numbers, we say so. A tool whose whole point is admitting when it didn't help cannot fabricate its own benchmark.