A B2B Team’s Review: Town

Sean Campbell
Authored bySean Campbell

This review is part of a larger series of LinkedIn newsletters titled The Human Side of AI: Cutting through the AI noise to show you how AI can be a powerful tool for your creativity, efficiency, and strategy.

Town is an ambient AI assistant that connects to your email, calendar, CRM, and messaging tools and works in the background. It triages your inbox, drafts replies in your voice, schedules meetings, and runs routines on a timer or a trigger. Each assistant, called a Townie, gets its own @town.com address, so you delegate to it the way you would to an executive assistant. Town launched publicly in June 2026, built by the former CTO of Plaid, and now lists 61 integrations.

On September 30 it shipped Town 2.0, which rebuilt the app around one ongoing conversation with your Townie and turned tasks into threads that teammates, and their Townies, can join.

Town is built for people whose week is mostly coordination. Sellers and operations leads get the most out of it.

What It Does Well

  • It works in the background: Routines run on a schedule or a trigger. A daily briefing, inbox triage, and a pre-meeting brief before every external call all happen without a prompt. You can also CC your Townie on an email thread and it picks up the task from there. That background mode is the main reason to buy it.
  • Delegation got simpler in 2.0: You open on a main thread and say what you need. The Townie decides whether to handle it there or spin it into its own thread that keeps running while you move on. You can reach it through Slack, WhatsApp, Telegram, or by voice.
  • It reaches the systems teams already run on: The list includes Salesforce, HubSpot, Pipedrive, and Attio on the CRM side, Gong, Fathom, Fireflies, and Granola for calls, and Asana, Jira, Linear, and monday.com for projects. If a tool is missing, you can connect any MCP server from a URL.
  • Approval is the default posture: Town says it never sends an email or calendar invite without your OK, and you decide per tool what it may do on its own.

How Each Role Puts It to Work

  • Salesperson: Before a discovery call, your Townie pulls the account from Salesforce or HubSpot, enriches the contact through Clay, and sends you a one-page brief. After the call, it drafts the recap email with agreed next steps for your approval and updates the opportunity. Pro includes 50 person research lookups a month, which covers a normal prospecting load. Heavy outbound teams will hit that cap.
  • Operations: Point it at a Front shared inbox and it sweeps the queue and drafts replies for approval. Set a Monday routine that pulls status from Jira or Asana into a project summary posted to Slack. Team plans share routines and skills, so one person builds a workflow and everyone installs it.
  • Researcher (secondary): Town handles the logistics around a study: the respondent-recruiting back-and-forth, interview scheduling, and a brief on each participant before the call. It does not code transcripts or ground findings in evidence.
  • Weaker fits: Marketers get Canva and LinkedIn drafting but no brand-voice controls beyond your own writing patterns. Finance can read Ramp, Brex, QuickBooks, and Xero, which helps with chasing receipts, but we would not let it touch numbers headed for a forecast without verification. HR gets Ashby pipeline tracking and nothing for evaluating candidates, which is the right call given the bias risk.

Where It Could Be Better

Everything the assistant does draws down credits, and overage bills per credit, from $0.044 on Pro to $0.030 on Team Plus. Town has added guardrails: the overage cap defaults to $0, a live tally shows usage, and an “Optimize credit usage” button audits consumption and suggests cuts. A team lead still has to estimate workload before budgeting, and browser and voice tasks add new ways to spend.

Organization-wide enforcement of no-training is available only as a term in a negotiated Enterprise agreement. On other plans, training is off by default but controlled by each user’s own toggle. For a security reviewer, that is a real gap.

Town also isn’t an analysis tool. It can summarize a meeting, but anything that needs a defensible conclusion belongs in a tool built to ground it.

Why Not Just Use ChatGPT, Gemini, or Claude?

Town runs on frontier models from providers like Anthropic and OpenAI. If you need a draft or a summary when you ask for one, a model you already pay for covers most of that. Town’s case rests on the background work: its own address, scheduled and triggered routines, 61 maintained connectors, approval gates, and shared team threads. It also changes the training default. Town says its agreements bar its model providers from training on your inputs and outputs.

Security & Compliance

Town is SOC 2 Type 2 audited, with reports, a recent penetration test, and a CASA Tier 2 assessment available under a self-serve NDA. It publishes a DPA with Standard Contractual Clauses and a sub-processor list. Data is hosted in the US on Convex and AWS with tenant isolation, encrypted with TLS 1.2/1.3 in transit and AES-256-GCM at rest. Town says it doesn’t copy or index your inbox, pulling only what a task needs, and no staff member has standing production database access. Google data is never used for training. Sign-in runs through Google or Microsoft, including Entra ID, with domain capture on Teams plans. We could not confirm SCIM provisioning, granular role-based permissions, or exportable audit logs. Raise those with the vendor before a rollout.

Data & AI Connectivity

Beyond the 61 integrations and custom MCP connections, Town can hand coding tasks to Cursor background agents, collaborate assistant-to-assistant on Pro and above, and use a browser to sign in to websites with logins you approve through 1Password. We did not find a public API or webhooks for building on Town itself.

Ratings

DimensionRatingRationale
Usability4.5 / 5The 2.0 redesign lets you start every request by talking to your Townie and leaves the organizing to it.
Power3.5 / 5Strong on briefings, follow-ups, CRM updates, and routines, but output still needs approval and review before it reaches a buyer.
Flexibility3.5 / 5Integration breadth is wide, but real depth reaches only sellers and operations, with light value for other roles.
Cost3.0 / 5A free tier and a $49/month Pro plan are reasonable, but credit metering makes team spend hard to forecast.

Best-Fit Roles

Strongest for Sales and Operations, useful to Researchers for logistics, and little here for Marketing, Finance, or HR.

Conclusion

Town earns its place in the work between meetings, not the work inside them. The 2.0 redesign makes delegation a one-sentence habit, and the connector list means it can act where your team already works. Keep approvals on for anything that leaves the building. The judgment calls on deals, priorities, and findings still belong to a person.

Last updated: 10/2/2026.

Home » AI Tool Reviews » A B2B Team’s Review: Town
Share this entry

Get in Touch

"*" indicates required fields

This field is for validation purposes and should be left unchanged.
Name*
Cascade Insights® will never share your information with third parties. View our privacy policy.