Twin1 AI does not want to automate your tasks. It wants to replicate you.
The San Mateo-based startup emerged from stealth on August 20 with a $20 million seed round co-led by Bessemer Venture Partners, Tribeca Venture Partners, and Aramco Ventures. Founded by Dr. Lewis Z. Liu, Tom Cahn, Huiting Liu, and Dr. Jonathan Budd, the company builds individual digital twins for knowledge workers. Not task-specific agents. Not workflow automation. Worker replication: the twin captures each person’s knowledge, judgment, work context, and communication style.
Liu’s background matters here. He previously founded Eigen Technologies, a document AI company that processed over $100 trillion in financial contracts before it was acquired by SirionLabs in 2024. He also served as a senior advisor to Linklaters, where he co-founded their Tactical Opportunities Group. That is not a coincidence — Linklaters is one of Twin1’s named customers.
The investor roster goes beyond the three co-leads. Angel backers include Wiz co-founder Roy Reznik, Notable Capital managing partner Hans Tung, and Dawn Capital co-founder Haakon Overli. But the most telling signal is Orrick, Herrington & Sutcliffe. Orrick is both a customer and a strategic investor. When a firm puts capital behind a vendor it is already deploying, that is a different kind of conviction than a standard procurement.
Law firms are the natural first adopters because they sell knowledge by the hour. If you can automate the communication patterns of a senior associate — the drafting, the client updates, the internal coordination — you are changing the economics of the billable hour. Customers including Linklaters, Orrick, Dechert, Customers Bank, and Aegis Energy report automating 30 to 50 percent of their communications work, according to company disclosures.
That number invites a hard question: what happens to the other 50 to 70 percent? And more pointedly, what happens to the workers whose entry-level communication is now being handled by software?
This is the junior gap. Historically, junior associates and analysts learned by doing the grunt work — drafting emails, summarizing meetings, handling the back-and-forth that senior partners did not have time for. If a digital twin absorbs that work, the apprenticeship model that professional services firms depend on starts to hollow out. The training pipeline does not just get more efficient. It gets shorter, thinner, and more dependent on software intermediaries.
Twin1 is positioning privacy and governance as the enabling infrastructure for all of this, not a compliance afterthought. The platform offers six layers of governance controls, model-agnostic deployment, and sovereign AI options ranging from SaaS to private cloud. It integrates with Slack, Teams, Outlook, Gmail, Google Drive, and SharePoint — the actual tools knowledge workers use daily. The enterprise MCP server and Twin Network coordination layer allow twins to share context across an organization while maintaining individual permissions.
The production ceiling question is worth noting. Only about 11 to 14 percent of enterprise AI agent pilots reach production, a commonly cited industry figure. Twin1 claims to be past that threshold with named enterprise deployments already running for over a year. Orrick CIO Wendy Butler Curtis called it “one of the most exciting developments in the practice today.”
The self-reported 30 to 50 percent automation figure deserves skepticism until these deployments scale beyond early adopters. But the direction is clear: enterprise AI is moving from automating discrete tasks to replicating the workers themselves. Law firms, where every hour of knowledge has a price tag, are simply the first place where that shift shows up on the balance sheet.