Most private capital firms under $10B have nobody who owns the data. Month-end takes weeks, the same borrower or fund shows up five different ways across the systems, and at some point a partner asks “what’s our AI plan?” and it lands on someone who already has a full-time job.
The 90-Day AI Program is that seat, filled. I come in as your Chief Data & AI Officer for one quarter: I meet every team, write the roadmap the partners can read, stand up one shared record, and put the first workflow live, measured against how long it takes today. I have built this function five times inside funds, lenders and insurers, and I am doing it now as Chief Data Officer at Altriarch Asset Management, a private credit manager.
What you have at day 30, 60 and 90.
By day 90, one workflow your team does by hand is running live and measured, the partners have a firm-wide roadmap they understand, and you are the person who made AI work at your firm.
Map
I meet every team and take stock of what you have: the tools, the data, the documents and anything already built. Each team gets a score on the AI Alpha Ladder. The AI seats, the policy and the training get set up, and your own team gets one quick win inside two weeks.
What you have at day 30- The firm-wide AI roadmap: what to build once and share, and the first workflow for each team, ranked by value, effort, data readiness and risk
- A plain-English data plan: what the firm needs, and does not need, on data lake, warehouse and lakehouse, with costs
- An AI Alpha Ladder score for each team
- A one-page AI policy, ready for counsel to sign
- A training session for staff, built on examples from their own work
- One working quick win
- The baseline: how long the first workflow takes today, in hours your CFO accepts
Build
I stand up the shared record, so last quarter’s numbers sit next to this quarter’s in one place. The first workflow gets built and tested, and we agree what “in production” means for your firm.
What you have at day 60- The shared data record, live, in your environment
- The first workflow live, with a named reviewer and a test set it has passed
- A runbook your team can operate without me
- A partner update: progress against the day-30 baseline
Prove
I measure the first workflow against the baseline, start the second team’s workflow on the same record, and write next quarter’s plan.
What you have at day 90- The results report: hours saved and accuracy, in numbers the CFO accepts
- The second workflow under way
- Next quarter’s roadmap and budget
- A partner presentation you can give yourself
One quarter, two days a week, led personally.
- Who
Led by Khurram Tehseen.The interviews, the roadmap and the build decisions are mine. Build work runs with a bench of senior data and AI engineers under my review.
- Conversations
The managing partner, the COO or CFO, and the head of each team.About an hour each in month one. After that, a weekly 30-minute check-in and a one-page update you can forward to the partners.
- Access
The systems those people complain about.Loan or deal system, fund administrator reports, the data room and the spreadsheets that hold it together. Nothing leaves your environment.
- Duration
Ninety days.Three monthly reporting cycles, so the change is measured on real work.
- Fee
A fixed monthly fee, quoted on the scoping call, agreed before work starts.No hourly billing and no change orders.
You can stop at day 30 and keep everything.
If the day-30 roadmap does not find value worth ten times that month’s fee, month one is free.
Everything produced stays in your environment. Nothing trains a public model.
Book 30 minutes to scope it →I work with two to three firms at a time. Senior people do the work, so the number stays small.
Three things the partners will ask for anyway.
For the partners, in your firm’s own nouns: lake, warehouse, agent, model, and which ones matter to you.
How the firm uses AI, written for the next DDQ and the operational due diligence call.
A question set for the AI section of your manager due diligence.
Who this is for, and who it is not.
For
- Private capital GPs and LPs under about $10B with no data or AI lead: private equity sponsors and their portfolio companies, private credit and specialty finance managers, family offices, endowments, pensions and funds of funds (LPs), and insurer-owned managers.
- The person a partner handed the AI plan to.
- Firms where a tool was bought and the spreadsheet never went away.
- COOs and CFOs who want the roadmap in their own nouns: the close, the DDQ, the quarterly pack, the capital call.
Not for
- Teams shopping for a tool demo.
- Firms that want a slide on AI strategy for the annual meeting and nothing underneath it.
- Firms without a senior sponsor willing to give five or six people an hour each in month one. Without that, the roadmap would be a guess.
Asked on most first calls.
How is this different from the free AI Alpha Ladder?
The ladder tells you your step in five minutes, with no call attached. The program is one quarter of me in the room with the people who run the firm: the roadmap, the shared record, and the first workflow live and measured.
What does the first workflow usually turn out to be?
For a private credit manager, compliance certificates recomputed from each credit agreement’s own definitions, or borrowing base certificates checked line by line. For a fund CFO, the month-end close reconciled against the administrator. For an LP, every manager’s capital account statements in one record. Month one decides which.
Do we need to buy a platform first?
No. Month one tells you what is worth buying and what is worth building. If a platform is the answer, we set it up and build around it. There is no platform of ours on the other side of the roadmap.
What do you need from us?
Five or six people for about an hour each in month one, read access to the systems they complain about, and someone who can approve the AI policy. Nothing is installed without your sign-off and no data leaves your environment.
What happens after day 90?
Most firms keep the seat on a monthly retainer: a workflow per team, the quarterly re-ranking, and the build work the roadmap calls for. Some hand the roadmap to their own team; the runbooks are written for that. There is no obligation to continue.
What if we are further along than we think, or not as far?
Then the roadmap says so, and you can still stop at day 30. Some firms learn the pilot they are proud of is running ahead of its foundation; others learn the boring work already done is most of the way there.