AI for Manager Selection and Fund Due Diligence

1. Picking Managers Is the Allocator's Core Job

For a family office or any allocator, selecting external managers is the job that drives most of the returns. Pick well across asset classes and the portfolio compounds; pick poorly and no amount of asset allocation saves it. Manager dispersion, the gap between good and bad managers in the same strategy, is wide, which makes selection one of the highest-leverage decisions an allocator makes.

It is also a lot of work. Building a view of the manager universe, running due diligence on candidates, reading the documents, analyzing the track record, and then monitoring the managers you back: each step is research-intensive, and most allocators do it with small teams.

AI accelerates the research and the document work so the allocator spends more time on what actually determines manager outcomes: the people, the process, and the alignment. It does not replace that judgment. This guide is about using it to do better, deeper manager selection with the team you have.

2. What AI Can and Cannot Judge

The boundary is sharper here than in most workflows, because what makes a manager good is largely qualitative.

AI can handle the data. Screen the universe, analyze track records, read DDQs and fund documents, and surface inconsistencies or things to ask about.

AI can support the process. Keep the diligence organized, draft summaries, and compare a manager against peers on quantifiable dimensions.

AI cannot judge the manager. Whether the team is talented and aligned, whether the edge is real and repeatable, whether the strategy still works, whether you trust them with the family's capital. These are the things that determine outcomes, and they are human judgments built on conversations, references, and experience.

The rule: AI does the analysis and the document work that informs the judgment. The allocator makes the judgment. A manager that looks good on paper and screens well can still be the wrong choice, and AI is no better than the data and the qualitative reality it cannot see.

3. The Manager DD Workflow AI Touches

Five stages, each with an AI angle.

Universe. Identifying the managers in a strategy worth considering.

Screening. Narrowing to a candidate shortlist against your criteria.

Due diligence. The DDQ, the fund documents, the track record, the references.

Decision. The investment committee or family decision to allocate.

Monitoring. Tracking the manager after you invest, watching for drift or trouble.

AI is strongest at the universe, screening, document review, and monitoring stages, the research and document work. The decision stage stays human. Note this connects to the other side of the table: the DDQ AI here is the buyer's version of what GPs face in our Fundraising and IR guide.

4. The Tool Landscape

The tools serving allocator due diligence.

Most allocators already have a data subscription. The fastest gains come from using its AI features properly and from applying document AI to the DDQs and fund materials that consume the team.

5. Building and Screening the Universe

Good selection starts with seeing the full set of managers in a strategy, not just the ones who found you. AI-assisted data platforms make that universe visible and screenable.

Preqin and PitchBook are the data backbones for manager and fund intelligence: who runs what strategy, with what track record, at what size, with what terms. AI-assisted search turns "show me lower-mid-market buyout managers with at least two prior funds in this geography" into a fast query rather than weeks of research, and helps screen the universe down to a credible shortlist.

This matters most for breadth. A small allocator team can only manually research so many managers, so without good tooling it ends up considering a narrow set, often biased toward the well-marketed. AI widens the aperture so the family is choosing from the real universe, which is where better selection starts.

6. DDQ and Document Review

Due diligence on a manager generates a mountain of documents: the DDQ, the PPM and fund agreement, audited financials, the track record, policies. Reading and cross-checking it all is the grind of manager selection.

DiligenceVault is built for the diligence side: a structured platform for issuing and analyzing DDQs, comparing manager responses, and tracking the process across candidates. AI on top helps analyze the responses, flag inconsistencies, and compare answers across managers or against prior years. For the fund documents, contract and document AI extracts the key terms (fees, liquidity, key-person provisions) and summarizes the long agreements.

The win is turning a slow, manual read into a structured, comparable analysis, so the team can see how managers stack up and where the questions are. The judgment about what the answers mean, and whether they are convincing, stays with the allocator. AI organizes and surfaces; people interpret.

7. Performance and Portfolio Analysis

Analyzing a manager's track record is essential and easy to do badly. AI helps do it more rigorously.

It can analyze the return stream, decompose performance, compare against benchmarks and peers, and read the underlying portfolio data the manager provides. For allocators already receiving statements from existing managers, document tools like Canoe read those reports into structured data, making both monitoring and the analysis of a manager's other funds easier.

The caution is the oldest one in manager selection: past performance is evidence, not proof, and a clean-looking track record can hide luck, a favorable period, or risks that have not yet shown up. AI makes the quantitative analysis faster and more thorough, which is valuable, but it cannot tell you whether the performance is repeatable. That remains a judgment about the manager's edge and process, not just the numbers.

8. Ongoing Manager Monitoring

Selection does not end at the allocation. Managers have to be monitored for performance, but also for the softer signals that precede trouble: style drift, team departures, asset growth that outruns the strategy, a change in tone.

AI helps by reading the ongoing reports and communications, summarizing performance against expectations, and flagging changes worth a closer look. It keeps the monitoring consistent across a roster of managers that a small team would otherwise track unevenly, paying attention to the squeaky wheels and neglecting the quiet ones until something goes wrong.

This feeds the family's broader oversight and the consolidated view, connecting to our Consolidated Reporting guide. The benefit is catching a deteriorating manager early, when you can still act, rather than at the annual review.

9. The Judgment That Stays Human

Worth stating plainly, because the temptation to over-trust a clean analysis is real. The things that most determine whether a manager succeeds are the things AI cannot assess.

Is the team genuinely talented, and will they stay hungry as they get rich? Is the edge real and repeatable, or a story fitted to a good run? Are their interests aligned with yours, and will they behave well when it matters? Do you trust them? These come from meeting the people, checking references, and the pattern recognition of an experienced allocator, not from a document analysis.

AI's role is to free the allocator to spend more time on exactly these questions, by taking the document and data grind off their plate. The best manager selection pairs thorough, AI-assisted analysis with deep human judgment about the people. Neither alone is enough, and the human half is the one that matters most.

10. Security and Data

Manager due diligence involves confidential fund materials, often under NDA, and the family's own allocation intentions. Both are sensitive.

Any tool that reads the materials must not train on them, must process them on vetted infrastructure, and must meet the family's confidentiality bar. Managers share their documents in confidence, and mishandling them damages relationships and reputation. The vendor framework is in our Security and Data Governance guide.

11. Where to Start

A practical sequence for an allocator.

First. Use your data platform's AI search to widen the manager universe you consider and screen it to credible shortlists.

Second. Apply a structured DDQ platform and document AI to the diligence grind, so manager comparison is consistent and fast.

Third. Set up ongoing monitoring that reads manager reports and flags changes, and protect the time this frees for meeting the people.

A Discovery Sprint can map AI across your manager selection and monitoring workflow, so the team does deeper diligence on more managers while keeping the human judgment where it belongs.