The order matters more than the ambition
Most AI work at investment firms fails in the same place: something impressive is built on top of data nobody trusts, and it quietly stops being used. We run engagements in the order that avoids that, even when it means a slower start.
A short diagnostic, paid, so that you may walk away from it. We follow the information rather than the org chart.
Two weeks sitting with the people doing the work. We trace a figure from the document it arrives in to the report it ends up in, and count the hands it passes through. What comes back is a candid map of your data estate and a costed plan — including the parts we think you should not build yet.
Foundations before cleverness — one definition of each number, and the lineage to defend it in front of a board.
The unglamorous half, done first. Ingestion from statements, capital calls and portfolio reporting; a model shaped around how you actually invest; reconciliation where sources disagree. When this is right, everything after it is cheap. When it is skipped, everything after it is theatre.
An assistive layer that drafts, extracts and flags, with a record of how every answer was arrived at.
Extraction at volume, memo and report drafting, monitoring that raises its hand rather than filing a report nobody reads. Built where the work already happens, so nobody has to remember to open a new tool. Every output is reviewable and every answer carries its provenance — a person keeps the pen.
It stays current with us or moves cleanly to your own team. We are not interested in being load-bearing forever.
Models change, vendors change, and your firm changes faster than either. We keep what we built current and review quarterly whether it is still earning its place. If you would rather run it in-house, you get the documentation and the enablement to do that — the IP was always yours.
What we hold ourselves to
- The people you meet are the people who build it.
- We work inside your environment and leave the IP with you.
- Where a simpler system beats AI, we will say so and build that instead.
- A fixed price for the diagnostic, and no surprises after it.
What we ask of you
- Access to the people doing the work, not only to the people describing it.
- One person who can settle a definition when two systems disagree.
- A decision-maker in the room at the end of each cycle.
There is no minimum term after the diagnostic, and no obligation to run the plan with us. If what we come back with is not worth doing, the right outcome is that you do not do it — and both sides would rather find that out in week two than in month six.