AI for Secondaries and Continuation Funds

1. Secondaries Went Mainstream

The secondary market is no longer a niche for distressed sellers. Lazard's 2025 Secondary Market Report put global secondary volume at about $233 billion, up roughly 53% on the prior year, and continuation-fund activity rose sharply alongside it. Continuation vehicles have become an accepted fourth exit route, next to a trade sale, an IPO, and a sale to another sponsor.

Growth that fast strains the way secondaries are underwritten, because the work does not scale with the deal count. Pricing a portfolio of fund stakes means understanding the companies underneath those funds, and that information is buried in capital account statements and quarterly reports from dozens of managers, each in its own format.

That is the bottleneck, and it is exactly the kind of document-heavy, repeating work where AI is strongest. The firms that can read more of the underlying, faster and at a fixed reference date, can look at more deals without losing rigor.

2. LP-Led vs GP-Led: Two Diligence Problems

Secondaries split into two kinds of deal, and they pose different diligence problems.

LP-led. An investor sells a portfolio of fund interests. The buyer has to value many funds and the many companies inside them, often with limited information, and the challenge is breadth: a lot of underlying assets to understand quickly.

GP-led. A manager moves one or a few assets into a continuation vehicle so existing investors can exit and new capital can come in. The challenge is depth and conflict: fewer assets, but the GP sits on both sides, so the diligence and the pricing have to be independent.

AI helps with both, but differently. For LP-led, it is about reading breadth at speed. For GP-led, it is about assembling an independent, bottom-up view that does not simply take the GP's marks at face value.

3. What AI Can and Cannot Do

The boundary, stated plainly.

AI can extract. Pull the underlying companies, cash flows, and marks out of capital accounts and quarterly reports into structured data.

AI can organize. Build the bottom-up portfolio view at a reference date and keep it consistent across many funds and formats.

AI can model. Lay out exit scenarios for an asset, comparing a trade sale, a sponsor-to-sponsor sale, and a continuation vehicle.

AI cannot price the deal. The discount or premium to NAV, the view on each manager, and the judgment on whether a continuation asset has another leg of growth are the heart of secondaries, and they belong to the investor.

The pattern is the familiar one: AI does the assembly, which is mechanical and verifiable, and people own the price and the decision, which are judgment with real money behind them.

4. The Underwriting Grind AI Attacks

Picture an LP-led deal: a portfolio of forty fund interests, each with twenty or more underlying companies, each fund sending a different capital account statement and quarterly report. To price it, someone has to read all of it and build a single picture of the underlying, at a single reference date.

Done by hand, that is weeks of analyst time spent typing numbers out of PDFs before any actual analysis begins. It also caps how many deals a team can look at, which in a competitive market is a real constraint on returns.

AI attacks exactly that grind. It does not make the investment judgment faster; it removes the data-entry wall in front of the judgment, so the team gets to the analysis sooner and can look at more of the market.

5. Capital-Account and Quarterly-Report Extraction

The core AI use in secondaries is extracting the underlying out of the documents that GPs send. Capital account statements, quarterly reports, partnership financials, and capital-call and distribution notices all carry pieces of the picture, in formats that differ by manager.

Canoe Intelligence is built for asset-level data extraction from exactly these documents and is widely used by LPs, fund-of-funds, and secondaries buyers. Chronograph is a portfolio-monitoring platform for LPs whose AI layer lets a team interface with the millions of documents they hold, from quarterly reports and capital accounts to LPAs. Accelex covers similar extraction ground. Together they turn scattered PDFs into structured portfolio data.

The verification rule applies with force here, because a secondaries price is built on these numbers. A figure extracted wrong from a capital account flows straight into the bottom-up value, so the figures that drive the price are checked at the source.

6. Bottom-Up Portfolio Analysis

Once the data is structured, the work is building the bottom-up view: what the underlying companies are, how they have performed, where the value and the risk concentrate, all anchored to a reference date so the picture is internally consistent.

Chronograph and similar LP-focused platforms unify data across fund commitments, secondaries, co-investments, and directs into one view of performance and exposure, which is the foundation a secondaries underwrite is built on. The same monitoring discipline that LPs use to watch a portfolio applies to diligencing one for purchase.

This is where breadth becomes manageable. With the underlying organized, an analyst can see the concentration, the vintage spread, and the names that drive the value across forty funds at once, instead of working through them one statement at a time.

7. Continuation Funds and Exit Scenarios

Continuation funds turn on a specific question: is this asset worth holding longer in a new vehicle, and at what price for the investors coming in and going out? Answering it means modeling the alternatives.

AI helps lay out the exit scenarios side by side: what the asset might fetch in a trade sale, in a sale to another sponsor, or held in a continuation vehicle with more time to grow. With the underlying data already structured, building and comparing those cases is faster, which matters when a GP-led process moves on a tight timeline.

The judgment stays human, and it is a sharp one. Whether the asset has another leg of growth, whether the price is fair to both sides, and whether the GP's view holds up are exactly the calls a secondaries investor is paid to make. AI frames the options; the investor chooses among them.

8. Pricing and the Discount

Pricing is where secondaries are won and lost, in the discount or premium to NAV that the buyer is willing to pay. It rests on the bottom-up value, a view on each underlying manager, and a read on where the market is clearing.

AI strengthens the inputs: a cleaner bottom-up value, a faster read of the underlying, and the scenario analysis behind the number. It does not set the discount. That is a market judgment shaped by competition for the deal, conviction in the assets, and the buyer's own return target, and it is the part no model should own.

The useful framing is that AI makes the analyst better informed at the moment of judgment, not that it replaces the judgment. A secondaries team with the underlying fully mapped prices with more conviction and less guesswork.

9. The Reliability Line

Two rules govern AI in a secondaries process.

Reliability. A secondaries price is a tower built on extracted numbers, so every figure that drives the bottom-up value is verified at the source. A confidently wrong mark pulled from a capital account does not look like an error; it looks like a slightly better or worse deal, which is the dangerous kind of mistake.

The price is yours. The discount, the manager views, and the decision to transact are the investor's judgment, documented for the investment committee. AI assembles the portfolio picture and frames the scenarios; people decide what to pay.

Inside these lines, AI lets a secondaries team look at more of the market and price with more conviction. Outside them, it industrializes a valuation error across a whole portfolio.

10. Security and the GP-Led Conflict

Secondaries diligence handles confidential fund and portfolio company data from many managers, often under NDAs and side letters. Any tool that reads it must not train on it, must run on vetted infrastructure, and must respect those confidentiality terms. Confidential work goes through enterprise tools or a custom agent in your own cloud, never a consumer account. The full framework is in our Security and Data Governance guide.

GP-led deals carry a specific issue beyond data security: the conflict of interest. The GP setting up a continuation vehicle is on both sides of the trade, which is why an independent, bottom-up view matters so much. AI helps here by making it practical to build that independent picture from the underlying documents rather than relying on the GP's own summary.

The diligence rigor of an LP buying secondaries overlaps with the work in our manager selection and fund due diligence guide, which covers assessing the managers whose funds you are buying into.

11. Where to Start

A practical sequence for a secondaries team.

First. Attack extraction. Pilot a capital-account and quarterly-report extraction tool against a real deal and measure the analyst hours freed before analysis begins.

Second. Put the underlying on an LP-focused portfolio platform that gives you a consistent bottom-up view at a reference date.

Third. Build scenario analysis on top, especially for GP-led deals, so the exit comparison is fast and independent.

A Discovery Sprint maps your secondaries workflow and shows where AI pays off first, from extraction to bottom-up analysis, with the controls that keep the numbers trustworthy and the data secure.