Private equity firms are adopting AI at very different levels of depth, and the distance between basic productivity gains and durable competitive advantage is becoming more pronounced. While many firms use AI to speed up research, drafting, and documentation, some are deploying more agentic systems that shape which opportunities partners see, when they see them, and how long those opportunities remain visible. The result is not just greater efficiency, but a structural shift in how deal flow is prioritized, revisited, and remembered over time.

At a recent IPEM webinar held in partnership with Datasite, Adam Ciborowski of RCP Advisors, Philippe Laval of Jolt Capital, and Henry Lindemann of Blueflame AI explored what this shift looks like in practice. A clear conclusion emerged: technology alone is no longer the differentiator. Advantage is increasingly tied to disciplined investment processes, high‑quality internal data, and firm‑wide adoption, with AI evolving from a support tool into an active gatekeeper to opportunity.

Where the industry stands: A spectrum of maturity

Adoption across the market remains uneven. Most firms today use AI as a productivity layer: drafting memos, summarizing diligence materials, and preparing presentations. These applications are valuable, but they are quickly becoming table stakes.

Others have embedded AI into specific workflows, including thematic sourcing, diligence tracking, and portfolio KPI monitoring. Only a minority, however, are operating at an agentic level: deploying systems that continuously monitor signals, surface opportunities or risks, and prompt follow‑up actions, while leaving final judgement and approval with investment teams.

Based on screening activity across hundreds of GPs, an estimated 30-40% are now using some form of AI‑enabled sourcing. That proportion is expected to rise quickly as tooling becomes easier to integrate and competitive pressure increases.

The opportunity: A persistent, always‑on deal pipeline

The defining advantage of agentic AI is scale combined with memory. In more advanced implementations, proprietary sourcing systems are already responsible for the majority of inbound deal flow, giving partners direct and continuous access to opportunities without relying on intermediaries or junior filtering layers.

Crucially, these systems do not forget. Companies assessed years earlier and passed on due to timing, maturity, or market conditions can be automatically resurfaced when underlying signals change. Over time, this creates a compounding pipeline where context accumulates rather than disappearing into past deal folders.

The value proposition is not simply more opportunities, but better prioritization. AI can narrow focus to the most relevant situations at any given moment, handing off precisely when human judgement is required. For firms overwhelmed by inbound volume, this means sharper triage. For lean teams, it offers broader market coverage without expanding headcount. 

The challenges: Data quality, adoption, and the '80% problem'

Despite widespread accessibility, meaningful performance remains difficult to achieve. Reaching acceptable accuracy is relatively straightforward; pushing beyond that, especially past the 80% threshold, is significantly harder. Achieving near‑complete reliability in a semantic, judgement‑based environment is an even greater challenge.

This creates a real risk of overconfidence, particularly when firms extrapolate from polished, conversational AI tools into mission‑critical investment contexts. Without careful design and validation, early success can mask fragility beneath the surface.

Beyond raw data quality, two structural factors consistently determine outcomes: the depth of a firm’s internal process data and universal adoption by senior professionals. The real edge often lies not in model sophistication, but in disciplined, firm‑wide usage. When AI is part of a daily investment workflow, it becomes embedded into decision‑making. When it is optional, it reverts to little more than an enhanced search tool. Changing long‑established partner behavior remains one of the hardest barriers to overcome.

Trends to watch: The build vs. buy debate is narrowing

Traditional distinctions between building and buying technology are beginning to blur. With foundation model licenses widely available, investment teams can increasingly prototype and adapt tools themselves. The strategic question is shifting from whether to build or buy, to where to do each.

A pragmatic approach is emerging: acquire data layers and infrastructure where they are commoditized and competitively priced, while retaining in‑house technical capability to drive iteration and learning. That internal feedback loop of connecting decisions, outcomes, and refinements is where sustainable differentiation now sits.

Importantly, the defensible moat is rarely the technology itself. Access is becoming universal. The moat lies in the underlying investment process and the ability to codify, modernize, and reinforce it continuously. Firms that have spent years capturing internal decision‑making data – every pass, every investment, every signal – are building a compounding advantage that is difficult to replicate quickly.

The horizon: AI at the investment committee?

Whether AI will ever have a formal vote at the investment committee remains an open question. The more immediate reality is subtler: AI already shapes outcomes by controlling which companies reach a partner’s attention in the first place. In practice, ranking is a form of decision.

Some firms are actively designing against this influence. One example involved rolling back AI‑generated pre‑investment memos that proved overly persuasive, replacing them with structured “red team” outputs that surface reasons not to invest. Rather than reinforcing conviction, the aim is to challenge assumptions, using AI to highlight risk, bias, and blind spots.

The broader implication is clear. Firms that treat AI as a feature to bolt on may see incremental gains. Those that treat it as a foundation starting with process, investing in data, and embedding it where investment work already happens are more likely to see lasting impact.

Insight. Innovation. Impact.

Continue the conversation about the tools reshaping the dealmaking landscape.