Findings / 53 respondents / Fielded July–August 2026
This survey was conducted in partnership with Addepar.
We asked asset owners how they use AI today, what is holding adoption back, and where they expect it to matter next. Explore the findings below.
Every chart below is built from these 53 blocks. One block, one respondent.
use general-purpose AI assistants — ChatGPT, Claude, Gemini. The next most-selected category, AI features inside existing investment platforms, was chosen by 25.
use AI daily in their own work. 8 use it weekly and 3 occasionally; none report using it at all.
expect their AI usage to increase over the next 12–24 months, 32 of them significantly. None expect to pull back.
How often respondents use AI personally, how far it has spread inside their organizations, which functions it touches, and which categories of tool they reach for.
Selected by no one: Not currently using AI tools
42 of 53 use AI daily. 4 of 53 report it fully embedded in core investment or operational processes.
Daily use is effectively settled — 42 of 53 reach for AI every day, and the tooling behind it is overwhelmingly general-purpose, with 52 using assistants like ChatGPT, Claude and Gemini against 25 using AI built into their investment platforms.
The gap worth watching is the one between habit and infrastructure. Most desks are past the question of whether people will use AI and into the harder work of embedding it: 4 of 53 describe it as fully scaled, while the largest group sits at “scaling.” The functions leading that shift are the operational ones — workflows, manager due diligence, portfolio monitoring — rather than the investment-decision core.
For an asset owner, the practical read is that individual productivity gains are already in the building; the durable advantage comes from moving AI out of the browser tab and into the systems of record, where it can act on your own data rather than whatever someone pastes into a prompt.
What respondents identified as obstacles to adoption, their single biggest concern, how their teams have responded internally, and the state of formal AI governance policy.
The question asked for up to three selections. 10 respondents selected more than three; all selections are counted here.
19 of 53 have a formal AI governance policy fully in place. 21 have one in development.
The barriers here are about trust and plumbing, not doubt — data confidentiality and integration top the list, while unclear ROI sits at the bottom with 6 of 53. That tells you the case for AI is largely made; what’s missing is the confidence to route sensitive portfolio and manager data through it and the connective work to make it fit existing systems.
Governance is visibly catching up but hasn’t caught up: 19 of 53 have a formal policy in place and 21 more are building one, which means a large share of daily use is running ahead of the framework meant to govern it.
The steer for asset owners is to treat governance as an enabler rather than a brake — a clear policy on data handling, model reliability and oversight is increasingly the thing that lets a team scale AI into fiduciary-grade work instead of keeping it parked in low-stakes tasks.
Where respondents expect AI to have the greatest impact, how they expect their own usage to change over the next 12–24 months, and what they said would most accelerate adoption.
The outlook is one-directional: 50 of 53 expect their AI usage to increase over the next 12–24 months and none expect to pull back, so the planning question is no longer whether to invest but how fast.
Where respondents expect that investment to land is telling — operational efficiency, portfolio monitoring and data management lead, ahead of investment research and decision-making. AI is being pointed first at the reporting-and-operations layer that consumes so much of an investment team’s time. And the single biggest accelerant named is unambiguous: 42 of 53 pointed to better integration with existing investment workflows, well ahead of training or ROI proof.
Read together with the barriers, the message for asset owners is consistent from both ends: what governs the next phase is neither enthusiasm nor evidence but whether AI can be wired into the data and systems you already run.
Alongside the multiple-choice questions, the survey asked two free-text questions. Respondents typed their own answers; the responses below are reproduced as submitted, grouped by question.
“Can you describe a specific use case where AI has delivered measurable value at your organization?”
Data normalization and processing have seen an incredible leap.On measurable value
Helping to build reporting APIs an integrations of investments. Really being a co-pilot to standing up our initial Addepar reporting instance. Bring enterprise level reporting to smaller boutique shop that we areOn measurable value
Investment analytics (raw data, data synthesis, presentation creation), accounting data reconciliation workflowsOn measurable value
manager due diligence, drafting IC decks and internal briefing memosOn measurable value
Evaluating manager documents to produce first cut diligence.On measurable value
Using AI to read through investor statements and updates, giving us a summary that we then pass along to clientsOn measurable value
Note taking in meetings, researchOn measurable value
We developed an AI process to perform the work a former operations employee was doing. The task entails collecting investment documents, renaming the PDFs to a standard naming convention, uploading/organizing them into a filing, and updating the values into our system of record. Besides collecting the document, the rest of the process has been automated saving hours of human work.On measurable value
meeting preparation. Synthesizes latest letters, pulls bull/bear cases on companies...what we need it to do is integrate with our historical notes in BipSync, generate this workflow over night in preparation for our meetings based on our personal calendars of upcoming meetings, look at e-mails to set context of meeting and history of introductions, etc.On measurable value
Due diligence, create diligence and reporting memos, portfolio management and risk analysisOn measurable value
We are using AI to proof legal documents.On measurable value
Decoding legal document languageOn measurable value
Building data models / IC Memos / Liquidity TermsOn measurable value
Nothing AI related has been automated so it is hard to say. But building a excel workbook to get a full picture of each of our managers has been helpful. Although it is still a wip and we are building it ourselves.On measurable value
The summarization of searchability of legal documents like PPMs and LPAs has saved hours of time.On measurable value
Streamlining investment committee reporting and presentationsOn measurable value
ODD preparation and identification of risk for new managers.On measurable value
Code gen tools are enabling idea generation and rapid prototyping from a diverse group including those without engineering backgrounds.On measurable value
Reconciliations between performance software and brokerage statementsOn measurable value
Our organization has seen the biggest gains from agentic uses - using recurring agents to compile summaries, emails, etc.On measurable value
Operational DD Questionnaire generationOn measurable value
AI generated portfolio dashboards have made Caissa data much more digestible.On measurable value
Speeding up the process of routine monitoring reports.On measurable value
Reconciliation for ABOR and IBOR. Writing memos for investments and other operational discussions. Assisting in custodian transition.On measurable value
Claude Code has automated our database health and monitoring. Also, we are able to build excel tools quickly.On measurable value
Editing e-mailsOn measurable value
I needed to parse an XML file with >5000 tags and turn it into a readable table to describe compliance rules to the traders. This was initially daunting. I described the output I was looking for and went back and forth with our AI assistant for about 20 mins and came out with a polished table.On measurable value
portfolio risk metrics and research consolidationOn measurable value
Broad based market research. Research primers on portfolio companies (getting from 0-50% knowledge quickly). Legal document summarization for investment staff.On measurable value
1. Transcribing meetings and producing instant meeting summary notes. / 2. Meaningfully accelerating the time to produce drafts of investment memos. / 3. Accelerated background research on specific industries / sub-sectors when evaluating dealflow / portfolio company activity.On measurable value
Yes, it's very helpful for training new hires, and for summarizing manager documents and meeting materials.On measurable value
Creating Investment Committee memos and reports.On measurable value
Risk Reporting, Scenario AnalysisOn measurable value
1. Hisotrical Data Management / 2. Building internal web app application / 3. Extracting HF exposure dataOn measurable value
Document retrieval and data extractionOn measurable value
“Is there anything about how your organization is approaching AI that you think is underrepresented in industry conversation?”
Bring enterprise level reporting for family offices to smaller boutique shop that we areOn what is underrepresented
Less frequent conversation about business continuity - not just how to integrate AI into workflows and processes, but how to ensure that other team members/future team members are also prepared to use/understand AI workflows.On what is underrepresented
I don't believe we are doing anything breath taking but some of the tasks and projects we have AI doing is extremely helpful to month end rec's, loan updating and investment reporting.On what is underrepresented
Have a business led approach than a IT led approach. The tone from the top is extremely important.On what is underrepresented
Connectivity. Without having all platforms that we use connected in some way any data we pull and use AI to analyze is only a partial picture.On what is underrepresented
Security is our primary concern, but that is already part of the industry conversation.On what is underrepresented
1. Data foundations first-- getting ALL the right data accessible to the right place at the right time without overloading and rotting context-- this is harder and matters a lot more in affecting outcomes than is appreciated. / 2. AI isn't a panacea. The right answers for some things are traditional deterministic processes or automation. You need to be willing to experiment, while accepting the fact that in some places "using AI" might actually cost you more than not, and is the wrong answer. The key is finding the areas it makes the biggest difference or has the chance of making the greatest difference and applying it there.On what is underrepresented
Finding an AI harness specifically designed for our needs.On what is underrepresented
Not truely underrepresented, but for independent research (investment due dilligence) it is much less reliable than workflows. We are working through how to make it better at research and continue to struggle.On what is underrepresented
We are actually looking at ROIs and not just spending to claim we are doing the "AI"On what is underrepresented
not yet - we're being somewhat cautious to roll out full agents with MCP connections. I have done this personally and seen the leverage.On what is underrepresented