Town's $1 Billion Valuation Prices a Network It Doesn't Charge For Yet

AI Capital · Two Deals, One Lens, part 2 of 2. Part 1 covers Otter's $320M repayable financing.

Bottom line. Town raised $108 million in equity at a $1 billion valuation, and it never has to repay that money on a schedule. Of the two instruments in this comparison, equity is the one that fits a product launched in June (moderate confidence), because there isn't yet enough customer history for a cohort-based financier to underwrite. Investors are pricing a future where each person's AI assistant learns them, works with other people's assistants, and becomes how teams, and eventually separate companies, coordinate work. Otter's financing assumes its past customers repeat; Town's equity is a bet on a product its current price list doesn't describe. Two gaps stand out. Town's pricing doesn't charge for the network, and its documented controls don't say how assistants at different companies resolve conflicts. The round buys time to close both.

Signal status. Developing. No retention data is public.

What's public, what isn't

On October 19, Town announced a $108 million Series B at a $1 billion valuation, co-led by Forerunner Ventures and Menlo Ventures. Town says it passed $1 million in ARR within weeks of its June launch, that paying customers have grown 4x since July, and that paying teams have grown 9x.

None of those multiples comes with a base. Town hasn't disclosed current revenue, starting customer counts, or retention. Four-fold growth on a few hundred customers and four-fold growth on tens of thousands describe very different companies.

You can't measure the size of the valuation jump either.

• Town began as a small-business tax service, launched in March 2025 with an $18 million seed led by First Round Capital.
• The assistant product arrived with a $55 million Series A on June 3, 2026, led by a16z, with Forerunner, First Round, Alt Capital, and Conviction participating.
• Neither earlier round disclosed a valuation.
• The new announcement doesn't say whether $1 billion includes the new cash.
• Tech Funding News puts total funding at about $181 million.

The founders account for part of the speed. CEO Jean-Denis Greze was Plaid's CTO. CPO Tony Vincent led applied AI products at Google, and before that founded Aspen, acquired by Google, and Sold, acquired by Dropbox. Equity investors in young companies weigh founder track records heavily. A financier repaid from customer revenue can't lean on one.

Why equity fits better here

Part 1 covered Otter's $320 million from General Catalyst's Customer Value Fund. GC describes that product as repaid out of revenue from each cohort, meaning each group of customers signed in a given month; Otter's own terms are undisclosed. Either way, that kind of financier needs years of evidence that cohorts behave consistently.

Town's assistant has been on sale for about four months. The tax business had different customers buying a different product. Whatever private data Town showed investors, not enough months have passed to see whether paying teams renew.

I'm not claiming Town was refused repayable financing. Nothing public says so. Moderate confidence: a cohort-based financier would have had very little to underwrite, and any offer would have been priced accordingly.

Equity puts the risk on the investors. If revenue arrives late, they own part of a company worth less than they paid, and Town owes nothing out of operating cash. What protections Forerunner and Menlo negotiated, such as liquidation preferences or board seats, isn't public.

That patience has limits, though. The $1 billion mark becomes the bar for the next round. Even if revenue has grown well past $1 million since summer, the valuation sits far above any revenue Town has described, so investors are paying mostly for what the product could become. I hold that direction at moderate confidence; the exact multiple can't be computed from public data. Speculative: Town needs evidence for the larger thesis within roughly 12 to 24 months.

What the $1 billion is buying

The lead investors published their reasoning.

Forerunner's Kirsten Green argues that an assistant grows more useful, and harder to replace, the more it learns about you. In her view, Town could become the front door to a person's digital work, built on habits and preferences other tools can't easily copy. Menlo's partners describe an assistant that works mainly through email, starts useful tasks without being asked, and eventually keeps a live picture of a worker's priorities across all their tools.

That's the single-user case. The larger bet is on a network. On Pro plans and above, assistants can work with each other. An assistant can also join email threads with people outside the company, including people who don't use Town, replying from its own @town.com address to schedule meetings and chase stalled threads.

Suppose enough people inside a company hand coordination to their assistants, and then people across companies do the same. At that point the valuable asset is the web of assistants that already know who owes what to whom. Moderate confidence: the $1 billion price requires the network. A single-user assistant would need a very large base of individual subscribers to justify that number on its own.

The price list describes a smaller product

Town's pricing:

• Free tier; Starter at $15 per month
• Pro at $49, Power at $99, and Power Plus at $199 per user per month, billed annually
• Paid plans include usage credits, with overage fees beyond them
• Team plans are per seat, with credits shared across the team
• Enterprise is custom
• Assistant-to-assistant collaboration unlocks at Pro
• Nothing is charged based on connections or coordination volume

Investors are paying for a network Town doesn't bill for. If the network grows, Town earns from it indirectly: more seats per company, upgrades to Pro for collaboration, more credits consumed as assistants do work for each other. That path is plausible. For now, though, the investor thesis and the price list describe different products. At Otter, the financing and the price list describe the same product, which goes a long way toward explaining why GC could underwrite Otter without taking shares.

Who is vouching for the product

Much of the public endorsement comes from investors who use the product.

• Green says that after a month she was running more than twenty automated tasks and sending 80% of drafted emails with light edits.
• Inc reported on Forerunner's internal use, including one partner's assistant asking another partner's assistant for information.
• Menlo published its investment case while it was both a customer and an investor, noting its assistants cost under $100 a month.

Weight this as strong evidence the product works and weak evidence that a market exists. Investors live in email and calendars and carry heavy scheduling loads, which makes them excellent early users and an unrepresentative sample. Their usage shows the product serves heavy coordinators. It says nothing about whether a large operations team keeps paying.

The controls decide whether the network can grow

The valuation needs assistants coordinating across teams and eventually across companies. Town's controls are the brakes on that coordination, so how tightly they're set bears directly on whether the network in the price can form. If you've shipped agents into enterprise deployments, you'll recognize the shape. These are Town's written policies; I haven't seen independent verification.

• Approvals. Per the approvals documentation, the assistant asks before making changes by default. Users can approve an action once, approve that action type for the session, or allow everything. Automated tasks run read-only, approval-required, or fully on their own.
• Owner check-ins. Per the external-email documentation, if someone asks the assistant for more than its owner authorized, it checks privately with the owner in a "sidebar." If the owner doesn't respond, nothing gets shared.
• Safe-action mode. The security page describes tighter restrictions when private data, untrusted input, and outside communication show up in the same task. It also states those protections are off in "Allow all actions" mode.

That last detail shows the tension most plainly. The mode that strips friction out of coordination also strips the protection against prompt injection, and every sidebar check-in is a point where coordination waits on a human.

Speculative: equity gives Town room to keep conservative defaults while it learns where autonomy is safe. A company repaying a financier out of near-term revenue would feel more pressure to loosen them. Investor terms aren't public, and equity doesn't guarantee patience.

Seat growth inside paying companies, one of the watch items below, runs through enterprise security review. Expect reviewers to press on team access: shared team connections reach whatever the connecting account can see, and fully deleting a departed user's account requires contacting support.

Where coordination stalls

Inside one company, the org chart settles conflicts between assistants. If two assistants both flag a request as urgent, the decision can go up to people who share a manager. Across companies, nobody shares a manager. The most vivid public example of assistants coordinating, the Forerunner partner whose assistant asked a colleague's assistant for information, happened inside a single firm where that structure exists.

Town's documentation covers local permissions and owner check-ins. It doesn't say who resolves a dispute between assistants at different companies, how competing urgent requests get ranked, or what one company's assistant may commit another company's people to. People outside Town just receive an email from an @town.com address.

If coordination stays inside companies, Town is a very good per-seat assistant priced like one. That's a solid business, and probably a smaller one than $1 billion implies. Speculative: the network thesis needs cross-company coordination to become routine, and that's the least specified part of the product today.

Portability works against the lock-in

The founders' October 19 essay states:

The next sentence says customers who leave can take that context, meaning what the assistants have learned, and use it elsewhere.

Set that next to Green's argument that accumulated learning makes the assistant hard to replace. If learned context really transfers, switching gets easier, and Town's lasting advantage has to come from somewhere else, most plausibly the network. If it exports as files another assistant can't turn back into the same behavior, the lock-in survives and the promise is narrower than it reads. Town hasn't published an export format or shown a customer switching.

I drew the same distinction in Three Bets on Who Keeps the Mortgage: owning your data and being able to keep operating after you leave a vendor are different things. Apply that test here. Ask what a departing customer's next assistant can actually do on day one.

Reading Town as an employer

Moderate confidence: a senior design leader at Town would own the decisions the valuation rests on: how much autonomy the approval modes grant by default, when an assistant interrupts its owner, what an assistant may say to people at other companies, and what happens to learned context when someone leaves. At a typical Series B, design scales a product whose pricing already matches what it does, and a miss slows growth. At Town, those autonomy and coordination choices help determine whether the network behind the $1 billion forms at all, and the answer will likely show at the next round, within 12 to 24 months. Take the role if you want your work to be the test of the thesis. Look elsewhere if you want to scale a product whose demand is already proven.

The product under each instrument

What to watch over the next six months

Town is valued at $1 billion within months of launching its assistant, with no public retention data behind the price. One deal isn't a pattern, so treat the idea that investors will price assistant networks this way as a hypothesis, and test it against the next round in the category. For Town itself, track three things:

1. Whether it discloses retention or an absolute customer count.
2. Whether seats grow inside companies that already pay.
3. Whether any customer describes assistants routinely coordinating with assistants at other companies.

If all three are still undisclosed at Town's next round, the network part of this valuation won't have been tested.