Many private market firms are facing two significant, increasingly opposed pressures: fee compression and the need to scale.
The first issue of fees is a result of the change in fees managers can command. Competition for the same capital has intensified, and investors are no longer willing to leave as much in managers' pockets. The industry's traditional 2% has been steadily eroding. Preqin reports the mean management fee for buyout funds raised in 2025 at 1.61%, the lowest on record, down from 1.85% in 2023.¹ Nothing in the current dynamic suggests the trend will reverse. This applies to Investment Managers, but is not confined to them. Wealth Managers are also increasingly competing on all-in costs to the client, and Fund Administrators are being asked to absorb more work at the same fee. Ultimately, these fees are borne by the investor, while Wealth Managers also need to demonstrate returns for their clients. When every party has a fee to cover, it puts further pressure on the client's net return.
The second pressure on scale comes from the wealth channel. In the funds I see, institutional fundraising typically involves twenty or thirty investors behind a fund. Opening the same fund to the wealth channel can increase that by considerably more than tenfold. Cerulli puts U.S. advisor allocations to less-than-fully-liquid private market strategies at $1.9 trillion today, heading to $3.7 trillion by 2029.² It is a different job. Most retail investments may be in the $50,000 to $100,000 range, and the challenge is not simply the number of investors. Regulatory and compliance requirements add complexity and liability, resulting in extensive manual steps that become increasingly difficult to manage as investor numbers grow. Onboarding runs at a different order of magnitude, and portfolio transparency at a different standard.
These pressures are not independent. The wealth channel makes the fee problem acute by adding a wealth management fee on top of the manager's fee. A manager now has to serve far more investors, meet a higher standard of speed and transparency, and manage the additional compliance and operational requirements that come with the wealth channel, all while the economics are getting tighter.
All while the economics are getting tighter. The right technology should support that growth through operating efficiency, a seamless investing experience and compliance guardrails. Ones that reduce both administrative burden and risk exposure.
Considering these market challenges, technology is a more efficient answer than headcount. It lets you grow the number of funds you manage without growing the team that manages them.
Here, the conversation tends to go sideways. Firms hear "technology" and reach for another tool. But another tool usually means another silo. A CRM, a financial system, an analysis tool, a valuation tool, and none of them talk to each other. The issue is less about the individual tools and more about what happens when the information they hold remains fragmented and difficult to use together.
Private markets are complex in ways public markets are not. That complexity lives in the data, and it arrives from every direction: your internal data, your administrators' accounting data, your investment data, your investor data. If it is not in Excel, it is in documents and PDFs. In my experience, nearly all managers are in this position, and we still see firms running billions of dollars in AUM out of spreadsheets.
Excel is a superb tool for modelling and testing scenarios. The problem comes when it becomes the place where the information underpinning your decisions ultimately lives. What you should not be doing is keeping the asset that defines your competitive position there, because data has become a competitive edge for a manager rather than a reporting obligation.
Agentic AI changes what firms can do with that data. The technology can now do far more with a firm's information, but without structured and governed data, much of that potential remains out of reach. These models are powerful engines, and an engine runs on what you feed it. Put the wrong fuel in, and it will not run properly; it produces more noise than power. That is where many firms are stuck. They can see what these models can do, but their data is not yet in a shape that lets them use them reliably. Every engagement we start at Investor Pointe looks beyond the “how do we use AI”, starting with the same question of "what condition is your data in?" The answer, without exception, is the same: it is everywhere.
The models most people know are large language models (LLMs). They are capable and general, but fundamentally reactive. Ask a question, get an answer. You are essentially using a very good browser.
An agent goes further. The model is only the brain, and it arrives knowing nothing about your firm. Everything that makes it useful is what we build around it. That includes the context of your organization, the tools it can reach and the harness it works inside. The easiest way to explain a harness is speaking about the hiring process. You give someone a job and then a job description. You tell them what they need to do, the limits they need to work within and where they fit in the business. A harness does the same for an agent, but also defines the numbers it should use. A person would build that calculation in Excel. We give the agent a trusted metric instead. It knows how that figure is calculated, where the information comes from and which sources it should use.
At Investor Pointe, those instructions carry ten to fifteen years of private markets operating experience baked into them. The model is the easy part, since anyone can access it. The context around it is what takes years to build. Our Portfolio Monitoring agent, for example, has been trained to understand how investor reports are structured and where the information is most likely to be. Once the data is extracted, it can apply predefined metrics to analyze the results and provide insights to the investment director.
That is what makes an agent proactive. It does not decide on your behalf, and I don't believe we are there, or that firms should want that. Its job is to put the right question in front of the right person at the moment it matters.
For example, let's say your latest report is generated, and revenue is down ten percent month over month, with nothing in the file to explain it. With an agent, it would be flagged immediately, along with other information about that asset and recommended next steps. Depending on your processes, the agent would stop there. The next steps are your call. The figure may simply be a reporting error, something may genuinely have happened at the company, or you may already know exactly why and have no need to notify them. Whichever it turns out to be, you make that call as soon as the information arrives, not when you discover the problem in a quarterly review two months later.
Investor onboarding works the same way. An Investor Engagement agent tracks where each investor is in the process and escalates as soon as it detects a problem, such as an investor stuck on a particular e-subscription step. It offers that investor help and raises the issue with the IR team at the same time, so a stalled subscription surfaces in hours rather than at the end of the month.
Distribution is where this gets most interesting. A fund is open, and a Distribution agent is monitoring engagement across the wealth channel: which platforms have the fund on their menu, which advisors have opened the materials, and which started a subscription and then stopped. The agent notices that an advisory firm that allocated to your last two vintages has not opened anything in six weeks, and, separately, that three advisors at the same platform all abandoned the subscription at the same document. Both go to the distribution lead as questions, not conclusions. One is a relationship conversation; the other may be a single fix that unlocks a dozen allocations, and neither of them shows up in a pipeline report.
We have built a team of expert agents, each focused on a different part of a manager's operation. One guides investors through the process of investing in a fund, another tracks asset performance, and another organises data so the rest can consume it reliably. They work much like a team of specialists, each with a defined role and access to the information they need.
The value is different for each part of the market. For Investment Managers, it means capacity. They can support more funds and more investors without a proportional increase in team. For Wealth Managers and Advisors, it means access. Private markets become more approachable, with much of the complexity and friction stripped out of investing in and monitoring private assets. It also gives clients the transparency they already expect elsewhere in their portfolio. For Fund Administrators, it means handling a larger volume of work without growing the operational team. For the end investor, it means knowing where they stand without having to ask.
Scepticism about AI in financial services is well founded, deserving of an answer rather than reassurance. Three things have to be true before any of this is worth doing. Firstly, your data must stay yours, which means the models in use cannot be learning from it. Secondly, every answer must be traceable, so you can walk back to the source of any single output. Third, the answers must be consistent.
That third one is the hardest. The known failure mode of these models is that they try to please you, producing something that sounds correct without actually understanding whether it is. In financial services, "sounds correct" is not an answer. It is either right, or it isn't, and the same question must return the same answer every time. If the underlying work hasn't been done, it won't, and then you are second-guessing every interaction, which is worse than where you started.
The common thread is the underlying data. That is why the data work comes first. You need everything under one roof, with governance, context and a semantic layer on top. We call the result agentic data, meaning data in a shape agents can use reliably.
What we see at Investor Pointe: one of our largest clients, a global private asset management firm, wanted to use the wealth of information sitting with their fund administrator. They also worried about how to do it safely. We brought that data under the manager's own roof, with governance over lineage and, just as importantly, over access, so sensitive information reaches only the people who should see it. Then we exposed it through an MCP interface, the emerging standard for how agents connect to data and tools. What their team previously could not reach became something they could interrogate directly, each person seeing exactly what they are entitled to see. Ad hoc reports that used to take weeks from the fund administrator now take minutes. The manager's response time has changed accordingly.
Today, the cadence of fund and portfolio reporting in private markets is quarterly. That cadence is already beginning to look slow. With evergreen and more liquid structures arriving through the wealth channel, monthly reporting and more timely access to portfolio information are likely to become a more reasonable expectation, and for some asset classes, faster still. While most operating companies are not reporting daily, expectations are changing.
Private markets are still seen, not unfairly, as something of a black box. Improvements in response speed, depth of analysis, and quality of investor communication are collectively pulling them toward the visibility investors take for granted in public equities and bonds. Beyond more frequent reporting, it is crucial to provide investors and the teams managing their portfolios with faster access to the information they need and greater confidence in what that information means. A reliable system of record means teams can identify issues earlier and respond faster. It also reduces manual work and creates a more consistent investor experience.
In our experience, the first call is usually fund administration data, the most accessible starting point and the area where firms most want more control and insight. Firms with a single administrator can move quickly; where several administrators are involved, the challenge is usually reconciling different data structures and integration layers into one consistent view. Then you work in waves, and each wave becomes usable the moment it lands. You don't have to wait for completion to start extracting value. Most firms see the first benefits within three months. The first fund or data set can become searchable and usable across the organisation, reducing the time spent requesting and reconciling information and giving teams faster access to answers that previously required manual work. As more funds and data sources are brought into the same environment, the value builds. More of the portfolio can be monitored consistently, reporting becomes easier to produce, and agents have a broader set of reliable information to work with. Timelines vary because consolidating third-party data is incremental work. Where we run the operation itself, in investor onboarding and engagement or portfolio intelligence, the value is more immediate. Agents can identify stalled investor journeys, surface portfolio issues and route the right questions to the right people without waiting for a manual review.
Much of what is marketed as AI in this industry is AI bolted onto existing software. The difference rarely shows up in a demonstration. It shows up in five questions:
The important thing is not simply whether a provider can answer these questions, but whether the answers give you confidence that the technology can be trusted in your operating environment. You need to be able to trace a number back to its source before using it in a decision. The same question should produce the same answer if you ask it tomorrow. Vehicles, investors and assets need to mean the same thing across your systems. Finally, you need to understand how your data is being used by the model, particularly where confidentiality and governance are concerned.
Private markets are being asked to do more with less. Fees are under pressure, more investors are coming through the wealth channel, and the operational burden that comes with them is growing. At the same time, the technology has caught up enough to do something useful with all the information firms already have.
That makes the data question much more practical than it used to be. If your data is spread across spreadsheets, PDFs, administrators and different systems, there is a limit to what any AI can do with it. Get that data into a consistent, governed environment, and you can start using agents to do some of the work that currently sits with your teams. They can find information, monitor changes, follow up with investors and surface issues that need attention.
You don't need to transform the whole operation at once. Start with the data that is most useful, prove the value, and build from there. The firms that do that now will have more options as the technology develops.
If you'd like to see how we've built this, including the specialist agents, the harnesses that govern them, and the data architecture underneath, we'd welcome the conversation.
Sources
Preqin, private capital fund terms data. Mean management fees on buyout funds: 1.85% (2023 vintage) and 1.61% (2025 funds, data through June 2025), the lowest average rate Preqin has recorded. Reported in Private equity management fees drop for the second year in a row, Preqin, October 2024, and in CNBC's coverage of Preqin's December 2025 report, January 2026.
Cerulli Associates, U.S. Private Markets 2025: Incorporating Private Market Investments into Model Portfolios. U.S. financial advisors currently allocate $1.9 trillion to less-than-fully-liquid private markets strategies, projected to reach $3.7 trillion by 2029.