The agents are in the investment committee — verification is the bottleneck

What changed

Ben Harrison, who leads industries at Intapp and founded DealCloud, told the Capital Allocators podcast that Intapp is repositioning its entire software stack to be AI-native, and that agentic AI — software that executes multi-step workflows without manual intervention — is where he sees the largest opportunity. His clients, he said, already have AI agents sitting in investment committees, testing what recommendations the agents make. The same firms are automating deal lifecycle management, from sourcing and underwriting through to committee proposals. m:…5381

Harrison also argued that vertical software keeps winning the wallet. DealCloud, he said, now serves 40 to 50 percent of the alternative investment market, and a majority of the top 500 to 600 general partners run its technology. Generic horizontal CRMs were built for standard operating models, he said, while dealmakers live in relationship-driven sourcing and multi-year deal structures. He cited research from Bain, McKinsey and others projecting private capital will grow past $20 trillion — and noted the asset class is democratizing toward retail investors, which multiplies the number of participants the software must serve. m:…0967

The expert network gets automated

The same automation is hitting primary research. Sagar Kadakia, CEO of Qualitate, said his firm replaces traditional human-led expert networks with autonomous AI interviewers grounded in proprietary datasets and client research frameworks. Each AI-led discussion runs 45 to 65 minutes without human intervention, he said, and the interviewer improves with every successive discussion by drawing on prior interviews. The company says it has grown its run rate tenfold in nine months and holds 300 percent net revenue retention after closing a $25 million Series A led by Insight Partners. m:…9969

Kadakia is explicit about where this goes: agents that autonomously scan data warehouses, spot information gaps, and launch interviews in real time — and he warns that when agents can launch thousands of interviews in minutes, token and compute costs will radically accelerate. His stated roadmap points at continuous intelligence infrastructure meant to make primary research faster, cheaper, and embedded rather than episodic. m:…9969

Verification becomes the product

That expansion of autonomous financial activity is creating a new infrastructure category: agent identity. Jonathan Awad, co-founder and CEO of Baselayer, told the M13 podcast that businesses are deploying agents to book travel and file applications, and banks need a way to distinguish an authorized agent from a bot. Baselayer, which says more than 2,300 financial institutions use its platform to verify and underwrite businesses, issues credentials that carry a business's verified identity to its agents — treating the agent as an extension of the business. m:…3056

Awad's pitch is that trust is transitive: if an institution trusts a business, and the business has no history of deploying bad agents, the institution can trust its agent too. The stakes are rising because instant settlement through stablecoins and FedNow removes chargebacks, he said, while breached personal data — about $20 a file on the dark web, in his telling — gives large language models cheap raw material for synthetic identities. The company raised a $35 million Series A in 2026 led by M13, according to the M13 post. m:…3056

What it means

This is my read, not a reported fact: the three moves form a stack. Harrison is automating the investment firm's internal workflow, Kadakia is automating the research that feeds its decisions, and Awad is selling the verification layer both need. Harrison himself flagged governance and security as the primary unanswered questions for firms deploying AI. For a VC, the pattern to track is which trust infrastructure gets standardized first — because whoever verifies the agent controls the toll. m:…5381 m:…3056