Analysis
Town, an AI personal assistant built for the enterprise, is raising at roughly a $1 billion valuation in a round led by Index Ventures, Newcomer reported on Aug. 27. The round size has not been disclosed.
The founding team is the reason this round exists at this price. Jean-Denis Greze was CTO of Plaid, the fintech infrastructure company Visa attempted to acquire for $5.3 billion before antitrust regulators blocked the deal in 2021. Tony Vincent was a product director for applied AI at Google. Andreessen Horowitz and Forerunner Ventures backed the company earlier. Town positions itself as the assistant for corporate work -- scheduling, follow-ups, inbox and coordination inside a company rather than a consumer's personal life.
The category is repricing in real time
Index co-led Instinct's $250 million Series B at a $2.5 billion valuation days earlier, alongside Benchmark, for a consumer assistant founded by 23-year-old Noah Shinn that is still in private beta. Pulse covered that round this week. One firm now has the consumer and enterprise sides of the same thesis. Martin, Lindy and Rox chase adjacent slices, and the incumbents are not idle: Microsoft Copilot, Google Gemini in Workspace and OpenAI's ChatGPT agents all ship assistant features into distribution Town does not have.
The honest risk
Enterprise assistants have a graveyard behind them -- x.ai's Amy, Clara, and a dozen scheduling startups that discovered calendar coordination is a feature, not a company. What is different now is that the model can actually execute multi-step work; what is not different is that the assistant sits on top of Google's and Microsoft's data, at their discretion. A billion-dollar valuation on a company whose core dependency is another company's API is a bet on switching costs that do not exist yet.
The enterprise framing is the interesting choice. Consumer assistants live or die on delight and word of mouth; enterprise assistants live or die on the security review. Town's founders come from Plaid, a company whose entire business was surviving bank security reviews, which is a more relevant credential than it sounds. Getting an assistant approved to read a company's calendar, email and internal documents requires SOC 2, data residency answers and an admin console -- work that has nothing to do with model quality and everything to do with whether the deal closes.
Pricing is the second unsolved problem. Per-seat pricing invites the comparison to Microsoft Copilot at $30 a seat, which is bundled into agreements companies already sign. Outcome-based pricing is cleaner in theory and nearly impossible to instrument for coordination work. TechCrunch's reporting on Instinct's consumer round shows the category's other half solving neither problem yet and being valued at $2.5 billion regardless -- which is a statement about capital availability more than about product maturity.
The talent signal is its own data point. A Plaid CTO and a Google applied AI product lead choosing to build an assistant rather than infrastructure says something about where operators with fintech-grade distribution experience think the next durable business sits. It also means the company will be judged on enterprise sales execution, a discipline neither founder has run at scale before.
The number to watch is seats per account after month six, which is where every prior assistant company has broken.