(each group a moderated session)
1. Teaching AI Your Firm's Data
Building the semantic layer: one shared definition of a fund, an LP and an IRR, mapped to where each one lives
Ask a plain-language question and the model has to land on the right records. That takes a semantic layer, and most firms don't have one: each system carries its own definitions, and no two agree. This session covers what to build first, who owns the definitions, and whether one layer can serve every team.
2. Building Private Credit's Data Layer
Loan tapes, borrowing bases, and getting to machine-readable data
Private credit scaled into asset-based finance and the wealth channel on documents and spreadsheets: loan tapes passed by email, borrowing bases rebuilt by hand, positions reconciled lender by lender. AI adoption among lenders doubled in a year; loan monitoring barely moved. This session covers where AI actually helps credit operations, and what shared data across lenders, agents and administrators would unlock.
3. Orchestrating a Firm's AI Agents
What routes, logs and permissions the agents, and when it earns its cost
A partner runs a CIM screener. An analyst runs a reporting macro. Once a firm runs more than a handful of agents, something has to route the work, log each action, and decide who may call what. That is the orchestration layer. This session covers what the minimum version looks like at a GP's size, and when it is worth the build.
4. Turning a Firm's Judgment Into AI Skills
Deal screening, diligence and the IC memo, written down so anyone, or any agent, can run them
A firm's edge is fifteen years of judgment: how to read a CIM, when a covenant matters. That judgment is now being written down as reusable skills. Writing them is easy. This session covers what to codify, who owns it, and how juniors learn the judgment when the skill does the work.
5. How Firms Check AI's Work
Who reviews the memos and the numbers, and how checking gets faster
AI will draft the memo and extract the numbers. Someone still has to check all of it, and checking is now the slow part: teams with heavy AI use produce twice as much and spend ~90% more time reviewing it. This session covers where the check sits in each workflow, what makes it faster, and which tasks still need human sign-off.
6. Redesigning the Org Chart: Who Owns AI at a GP
Forward deployed engineers, AI champions, and getting use cases to production
Most enterprise AI pilots show no measurable return, and the failures sit in messy data and workflows, not the model. The vendors' answer is the forward deployed engineer, embedded at the client, and postings for the role are up more than 700% this year. This session covers which roles a GP needs at its size, and where they report.
7. How AI Is Changing the Work of Fundraising and IR
What works today across the raise, from diligence responses to LP coverage
More than half of LPs say committed but uncalled capital is keeping them out of new funds, and fewer funds are closing. Raise teams are adopting AI under that pressure, and most fundraising and IR professionals are not AI experts: the tools that work get built together with the firm's operations and technology teams. This session covers what is working in the raise today, and which steps still need human review.
.png)
.jpg)
