Town

• Having a Townie sort the inbox each morning, draft replies in your voice and leave them on the thread for review
• CCing your Townie on an email with an outside partner to find a meeting time across three time zones and send the invite
• Getting a brief before every external meeting that pulls the last call notes from Granola, CRM history from HubSpot or Salesforce, and recent Slack threads
• Logging call notes and next steps to the CRM after each demo and drafting the follow-up email for approval
• Rolling out a team call-prep routine once so every rep's Townie preps calls with the same updated battle cards
• Texting your Townie from a hallway for talking points before a meeting, or approving a pending send by replying yes

Town's bet is that the useful unit is one assistant per person, connected to that person's inbox and calendar, rather than one bot for the whole team. The documented pieces support that: a learned profile and wiki, routines with real triggers, CC-able email threads that include outsiders, and team features that share setup without merging everyone's data. The approval model is unusually granular, which matters for something that can email clients and book meetings. The main cost of entry is predictability. Credits are the meter, per-task costs are not published, overage is off until you raise the cap, and the features that make Town more than a chat assistant start on Pro. Use the 7-day Pro trial on the routines you actually need, then check the per-routine credit breakdown on the Plan page before rolling it out to a team.

About Town

Town starts with one person and one assistant. You sign in with a Google or Microsoft account, connect your email and calendar, and name your Townie, which gets its own @town.com address and builds a profile and a nightly-updated wiki of your people, projects and writing voice from what it can see. From there you reach it in the web app, the iPhone and Mac apps, Slack, email, WhatsApp, Telegram or plain text messages, and ask for work the way you would ask a colleague: catch me up on my inbox, find 45 minutes for the three of us, draft a reply, build a deck from these numbers. Repeat work becomes Routines, plain-language jobs that run on a schedule, on a new email, before a meeting or from a webhook, such as the stock Auto-inbox, morning briefing and pre-meeting brief, plus a public library of role-specific routines for sales, recruiting, finance and investors. What separates Town from most personal assistants is the network around each Townie. You can CC your Townie on any email thread, with people inside or outside the company, and it replies from its own address and checks privately with you before sharing anything. You can add it to a group text where it trades availability with other people's Townies, and since September 2026 it can use a browser to sign in to sites, book and buy, stopping at checkout for approval. With its October 2026 Series B announcement, Town said Townies can now work directly with each other inside a company, with customers, partners and candidates over email and text, and through team-wide shared workflows. Teams get shared integrations, installable routines, background routines owned by the team, shared skills and access rules, with pooled credits on paid Team plans. Control is layered: sessions default to asking before any change, routines can be read-only, approval-required or autonomous, and each tool can be set to ask every time or always allow. On the security side, Town says it is SOC 2 Type 2 audited, hosts data in the US on Convex and AWS, keeps AI training off by default and never trains on Google data. The tradeoffs are mostly about metering and reach: paid plans are measured in credits whose per-task cost is not published, the useful automation (event triggers, unlimited routines, assistant-to-assistant work) starts on Pro at $49 a month billed annually, sign-in is limited to Google and Microsoft accounts, messaging apps are not reliable for proactive alerts, and there is no self-hosted option.