The best of what exists today.
Our agents are built on Anthropic's Claude foundation models — the most safety-aligned large language models commercially available — and orchestrated via the Model Context Protocol (MCP), the open standard for structured tool use and agent-to-system communication. Inter-agent coordination follows Google's Agent-to-Agent (A2A) protocol, enabling typed, traceable handoffs between autonomous systems. The architectural blueprint is MIT Media Lab's NANDA framework — defining how agents discover, authenticate, federate, and interoperate across trust boundaries.
These are not proprietary claims. They are open standards, peer-reviewed research, and publicly auditable protocols. We chose them because building AI for private capital markets demands the same rigour that governs the capital itself.
This commitment to excellence is the founding principle of the Forum Foundation — TABULARUM's initiative to promote best practices in artificial intelligence, foster open research in agentic systems, and set the standard for responsible AI deployment in financial services. The agents below are the first implementation of that vision.
Built by professionals,
powered by AI
Our agents are not generic chatbots. Each one has been designed and trained in collaboration with fund lawyers, investment analysts, CFOs, and operations directors with decades of experience in private capital markets. They encode real-world expertise — not just language models — into every document they draft, every analysis they run, and every workflow they orchestrate.
Meet the agents
Agents that collaborate
No single agent works in isolation. When a GP launches a fund, Gaio drafts the legal documents, Mila sets up the cap table, Clara manages the data room and investor outreach, and Lucio screens the pipeline. Every handoff is automatic. Every action is logged. The result is a fully orchestrated operation that would normally require a team of 20.