Key Takeaways

  • Ben Harrison, President of Industries at Intapp and former DealCloud leader, argues that simple natural language search fails to build a competitive edge in private equity origination.
  • Top-performing deal teams combine global public data with licensed third-party subscriptions and internal institutional records to map broad market sectors across thousands of potential targets.
  • Contextual software agents now ingest historical call records, previous pitch decks, and relationship histories to execute programmatic outreach strategies against target executive teams.
  • Origination advantage has shifted away from commoditized market databases and toward secondary and tertiary proprietary records that exist only inside the firm.
  • Harrison organizes this systematic origination workflow through the Three-Layer Data Foundation for Systematic Sourcing.

The Three-Layer Data Foundation for Systematic Sourcing

According to Harrison, investment teams face a major shift in how they evaluate target markets: “On the sourcing and origination side, it has become incredibly fast and a heck of a lot easier to map the market, research the companies, develop the hit list, and develop strategies against those particular assets.” Yet standard database queries no longer create proprietary deal flow. As Harrison explains, “Now, instead of researching that and searching that or just doing NLP question and answer with AI, you can build playbooks and agentic coworkers that can ingest all of that and pursue a strategy across all of those assets.”

Winning firms separate their sourcing engines into structured tiers of external and private knowledge: “It's the secondary and tertiary layer of that data that's forming some of the most thoughtful and thorough sourcing and origination techniques.” Harrison structures the pipeline through four distinct components:

Layer 1: Public Market Data

Start with global, publicly available company and market data to map the broad universe of potential assets.

Layer 2: Proprietary Third-Party Feeds

Add licensed external financial databases, market feeds, and industry-specific intelligence subscriptions to enrich company profiles.

Layer 3: Internal Proprietary Relationship Data

Integrate the firm's unique institutional history, including calling efforts, previous deal interactions, email relationship intelligence, and internal documents.

Layer 4: Agentic Execution Playbooks

Deploy context-aware AI agents across the integrated data foundation to filter assets down from thousands to tens, matching asset economics against the firm's investment criteria to generate tailored management pitches.

When This Works (and When It Doesn't)

Harrison notes this architecture works when private equity and investment banking teams run systematic origination across thousands of potential platform or add-on targets. The model excels in fragmented sectors like healthcare services, vertical enterprise software, or specialized industrial distribution where fund returns depend on outbound sourcing and relentless add-on acquisitions.

The framework breaks down when funds pursue concentrated, large-cap buyout strategies driven by investment bank auctions. When only ten potential targets exist globally for a given mandate, algorithmic filtering adds little value over direct founder and board relationships. The model also fails when a sponsor suffers from poor internal record-keeping. If deal teams refuse to log relationship touchpoints or partners keep interactions siloed in private inboxes, AI agents run on fragmented, distorted inputs that degrade outbound outreach.

Why It Matters

Origination has changed from an individual partner rolodex game into an institutional data race. Public registries, scraped web metrics, and commercial market databases have become universal commodities. When every private equity shop accesses identical company registries, external market feeds yield zero pricing or access advantage. Value accrues entirely to sponsors that capture proprietary relationship history and run programmatic execution against those records.

This dynamic widens the performance gap between institutional platforms and mid-market boutiques. Multi-fund sponsors with decades of structured interaction data can deploy agents to identify which partner met an executive three years ago, what terms were discussed, and why market timing aligns with an outreach effort today. Sponsors operating without integrated internal data layers will find themselves bidding late in competitive auctions against rivals who initiated proprietary conversations months earlier.