This interview was originally published on PE Hub.
Juniper Square’s Brandon Rembe explains why private fund managers need to rebuild their operating model around AI, and how to take the first step.
Agentic AI promises to transform how private fund managers operate, from back-office processing to portfolio operations to fundraising. Yet embedded legacy systems and ways of working often limit AI use to basic tasks like drafting emails or building slides. Brandon Rembe, chief solutions officer at Juniper Square, describes the shift fund managers need to make to harness agentic AI’s full potential.
Q. Why are private fund managers’ current operating models incompatible with AI?
Rembe: Any knowledge-based, data-driven industry is grappling with how to implement AI; consulting, accounting, legal, and private markets GPs are all facing the same question. The default response is to plug this new technology into existing workflows, but that’s like wiring an electric motor into a steam-powered engine. To capture the full potential of AI, managers have to rebuild their entire factory around it.
What that rebuild unlocks is the ability to outsource intelligence at scale and at a fraction of the cost it used to require. Using AI to polish an email or format a slide amounts to running spellcheck on Word. Managers can now use agents to run end-to-end workflows across investor onboarding, fundraising, quarter-end reporting, and data review and verification – with a human expert in the loop to check the output.
Q. What do managers need in place to adopt agentic AI?
Rembe: Whether you’re building agents in-house or outsourcing to an operations partner, AI cannot work without context, and context is data. The good news is that AI itself is very good at creating some context from disparate documentation and data. But managers need to ask whether they have the right historical reports, the right data pipelines, and the connected infrastructure to keep everything current.
The structural challenge in the private markets is that there is almost no standardization in how data flows. From portfolio companies to property managers to fund accountants, every firm does things differently. On top of that, a significant amount of operational knowledge lives in individual employees’ heads and inboxes. Systematizing that tribal knowledge at scale, as close to real time as possible, is the hard-to-build foundation every firm has to lay.
Q. How can managers capture that institutional knowledge?
Rembe: At Juniper Square, we call it the “Doug problem.” There’s always a partner—somehow always named Doug—who is a rockstar in client meetings but never takes notes, and then someone else has to chase him down afterward to reconstruct what was said. The easy fix is to let an AI agent join those meetings. Ideally, that AI automatically feeds everything back into the firm’s CRM so everyone in IR has the full picture, no matter who is physically in the room.
The harder tribal knowledge is the twenty-step quarter-end review that lives only in one person’s head. That’s where managers have to put in the work. Sit with AI for ten minutes, walk it through your process, and have it draft a workflow diagram. It will hit 80 percent accuracy shockingly quickly, and you can refine from there. Think of it like training a new hire; the more you coach it, the smarter it gets.
Q. AI is evolving fast. What’s the first step that won’t overwhelm a firm?
Rembe: Pick one small end-to-end workflow that causes real pain. Ask the person doing the work today what inputs they need, feed that into an agent, and watch the result. See what worked, see what didn’t, and learn by doing. It’s a great
way to see the potential.
But don’t assume you have to build everything yourself. When the internet arrived, everyone thought they needed to build their own website and hire their own developers. Almost no one does that anymore. The same pattern
will play out with agentic AI.
GPs probably shouldn’t try to hire armies of agentic developers anyway—they’re hard to find, expensive, and tend to want to work for the hyperscalers.
More than 40 percent of PE GPs have an AI strategy for their own businesses. But the execution question—who builds it, maintains it, and ensures it stays compliant—is one most GPs are not equipped to answer alone, nor should they be. For most GPs, it’s a partner selection question.
The demands of a GP’s C-suite are changing. The new mandate is to find partners that can absorb the AI burden—operational load, compliance posture, talent bench, and infrastructure –on the firm’s behalf.
Q: What impact is AI having on how firms think about work itself?
Rembe: The most common refrain I hear is: “What’s the best tool for XYZ task?” In reality, every exec should be asking themselves: “How does this person do their job today, and what would it take for an agent to do that work in the future?”
It shifts the focus from outputs to outcomes. An output is an email or a Word document. An outcome is the ability to respond to any due diligence questionnaire with hyper-accuracy within a 24-hour service level agreement. Or raise a billion-dollar fund at twice the speed with half the personnel.
The GPs moving decisively are those whose operations partner is already running AI‑enabled workflows against the firm’s data, with the governance layer, services team and accountability model already in place. Almost every software vendor has added AI to their product in the last 18 months. And most of them cannot meet the standard that private markets GPs should be holding them to. The test is not whether a tool has AI features. It’s whether the operations partner behind the tool has built something a GP can actually rely on for investor‑facing, fiduciary, and compliance‑bearing work.
Q: How should firms use agents day-to-day?
Rembe: Onboard an agent the way you’d onboard a new employee, with the same level of oversight at the start. Over time, the level and type of oversight will change as agents build trust within the organization.
The software interface itself is changing, from software-as-a-service to service-as-a-service. Managers are moving away from buying more checkboxes and picking lists for their staff to maintain and toward buying finished outcomes.
And the new layer on top of all of it is verification. For agents to gain trust in an organization, they have to show their work—the documents they pulled, the calculations they ran, the Python code they used, the discrepancies they flagged.
Think of it the way we think about ordering a pizza. In the US, I can track a $9 Domino’s order from oven to doorstep. That’s the level of auditability a CFO should expect on quarter-end processing or on a KYC/AML check. Once that transparency is in place, verification gets lighter over time. The first time someone booked a flight on Expedia, they called the airline to confirm the flight. By the fourth time, they trusted Expedia. The same trust curve applies to agents.
Q: And firms that resist?
Firms can’t afford to bury their heads in the sand. They will run less efficiently, which means higher fees and lower returns. They will spend their time on back-office processing rather than on the things that drive performance. The first instinct might be to wait it out, but the only thing to do is to embrace it and get started.