Building the governed context layer for enterprise AI with Twin1

The most valuable knowledge in a professional organization lives within the minds of its people. A law partner carries years of deal precedent and judgment into every engagement, and the third-year associate knows which paralegal turns a closing checklist around overnight or which specialist to pull in before a question becomes a problem.

For centuries, firms have moved this context through hierarchy, meetings, and apprenticeship, because human memory was the only place it existed. The promise of enterprise AI has fallen flat thus far as tools have consisted of copilots and search boxes that can only retrieve whatever might be saved to a file store. These tools carry no context about the firm itself, unable to tell what one colleague is permitted to share with another, which materials are highly sensitive, or how relationships work across the organization. In legal and financial institutions where confidentiality and security are paramount, these governance concerns make adoption significantly harder. There is a subtler cost as well. When a company routes everything through a centralized assistant, differentiated expertise gets aggregated and averaged into generic answers, a failure mode researchers call knowledge collapse.

We partnered with Twin1 because it starts where the knowledge lives, with the individual. Every knowledge worker gets a digital Twin, a live model of their expertise and judgment that operates inside Slack, Teams, and email. A Twin fields routine questions on its owner's behalf and mirrors their privacy preferences down to the specific topic and colleague asking. Networked together, these Twins form a governed context layer for the entire organization, one that people and AI agents can query without ever bypassing the person who owns the knowledge. 

Each new Twin expands what the network can answer. The tacit context it captures exists nowhere on the public internet, beyond the reach of any frontier model, and much of it never even lands in the company’s own file stores. We believe this layer grows more valuable as agents take on real work inside companies, since agents need the same permission-aware access to context that humans do. As Twin1 launches out of stealth, the company is already in production across businesses in legal, financial services, and energy.

Our deepest conviction in Twin1 lies with the team. The founders have worked with one another for more than a decade. Lewis Liu, Tom Cahn, and Huiting Liu built Eigen Technologies together, whose document AI processed more than $100 trillion in contracts for half of the world's largest banks and a fifth of the AmLaw 100, while Jonathan Budd spent those years building J.P. Morgan's largest RAG system and co-founding the encrypted storage protocol Peergos. They all deeply understand how regulated enterprises evaluate, procure, and trust software after spending years inside the rooms where those decisions are made.

We’ve known this team since their Eigen days and watched the Twin1 thesis take shape from its earliest conversations. Now, we’re proudly co-leading Twin1's seed financing alongside our friends at Tribeca Venture Partners and Aramco Ventures, and look forward to what Lewis, Tom, Huiting, Jonathan, and the Twin1 team build from here. Want to explore Twin1 for your enterprise's AI systems? Book a demo here.