Key Takeaways

  • Ben Harrison separates frontier models like Claude and ChatGPT from firm AI designed to execute institutional workflows.
  • Over the last 18 to 24 months, agentic automation enabled alternative asset software to move from prompt-and-response queries to autonomous execution.
  • Workflows with rigid compliance and sequence rules, including deal lifecycle tracking, LP onboarding, and fundraising, represent the highest-yield targets for firm agents.
  • Junior investment staffing models face pressure as context-aware agents take over repetitive administrative screening and multi-step deal procedures.

Frontier Models Excel at Ad Hoc Tasks While Firm Procedures Require Memory

Most private equity professionals interact with artificial intelligence through general-purpose chat interfaces. Analysts use tools like ChatGPT or Claude to summarize call transcripts, parse regulatory filings, or generate starter financial models. Harrison points out that while frontier models excel at these isolated tasks, they lack context about how a specific private equity shop operates.

Every investment firm runs on custom procedures. A buyout fund may require an associate to log every sponsor interaction, run conflict checks against active pipelines, and route early diligence memos to specific sector leads before an investment committee review. A public chatbot cannot complete that chain. As Harrison explains: “Where our agentic AI lives inside of these organizations is on the execution of the firm's procedure.”

Firm AI ties large language models directly into an institution's historical database. When the software retains institutional memory, it stops behaving like an external research assistant and begins functioning like an internal operator. “It is that procedure that a context-aware agentic AI agent can run through, learn, process, and automate where you can get an incredible outcome,” Harrison says.

The Shift to Autonomous Multi-Step Execution

The software industry spent years building relational databases and pipeline trackers like DealCloud. Deal teams still had to manually input interactions, tag contacts, and move deals through internal stage gates. Harrison argues that the technical baseline altered rapidly.

“What happened in the last 18 to 24 months with the agentic automation is the biggest step we've ever had in the ability to take the software and run it autonomously for the organization,” Harrison notes. Instead of waiting for a user to trigger a search, agentic software detects an incoming communication, assesses its relevance against pipeline data, drafts the appropriate records, and flags exceptions for senior partners.

This shift directly alters mid-office and back-office operations. In fundraising and LP relationship management, firms spend weeks managing information flow, answering due diligence questionnaires, and onboarding new capital commitments. “Today, the super-complex workflows, the deal life cycle management, fundraising and LP management, each of the different areas of the business that have a procedure behind it, we can drive speed and efficiency at a level that we haven't seen before,” Harrison says. When software independently runs multi-step administrative processes, it cuts the coordination drag that normally expands as asset managers add strategies.

Why It Matters

This shift signals that enterprise value in vertical software will accrue to vendors that hold proprietary operational workflows rather than generic AI wrappers. For private equity managers, capital deployment speed and lean overhead will separate top-tier operating margins as firm-specific agents reduce reliance on armies of junior administrative staff.