Capsa AI Raises $18M Series A to Build Agent Infrastructure for Private Capital Firms
Capsa AI, a London and New York-based AI knowledge platform for private capital, closed an $18M Series A co-led by TX Ventures and Pivot Investment Partners. Bek Ventures, Antler, Outward VC, and Cornerstone VC also participated, alongside angel investors including Indeed co-founder Paul Forster. Total funding to date stands at $20M.
What Capsa AI Does
The platform indexes a fund’s entire data estate, pulling from CRM systems, Outlook, SharePoint, and proprietary sources into a single knowledge layer, according to FinSMEs. That layer captures how a fund thinks and operates: every memo, conversation, and decision the firm holds. Agents then execute workflows across the fund lifecycle, from deal sourcing and due diligence through portfolio monitoring and back-office operations.
The architecture is model-agnostic, SOC 2 Type II certified, and offers single-tenant deployment with regional data hosting. Capsa AI does not train on client data, a non-negotiable requirement for regulated financial institutions handling confidential deal and portfolio information.
Who Uses It
Several of the world’s largest private capital firms are already using the platform, with additional enterprise deployments rolling out across the US and Europe, The Paypers reported. The team currently numbers 14 people across London and New York.
Founded by Danyal Oezduezenciler and Callum Downie, Capsa AI plans to use the Series A proceeds for US market expansion, engineering and go-to-market hiring, and further development of its agentic and indexing capabilities.
Why $18M Matters for Agent Infrastructure
The raise is notable for what it signals about agent adoption in regulated industries. Private equity, venture capital, and hedge funds are among the most data-intensive and confidentiality-sensitive sectors in finance. A funded startup building agent infrastructure specifically for these firms suggests the compliance and security barriers that slowed initial enterprise AI adoption are being solved at the infrastructure level.
The single-tenant, zero-training-on-client-data approach addresses the core objection that has kept many financial institutions from deploying agentic systems: data leakage risk. If Capsa AI’s model scales, it establishes a template for how agent platforms can operate inside industries where information barriers are legally mandated.