Smith vs Town: an assistant for you, or coworkers for your team
· Avi Agola, The Seattle Compute Company
Town and Smith both give you AI agents with their own addresses, so people lump them together. They're built for different jobs. This page lays out where they differ, and where Town is the better choice.
Everything we say about Town comes from town.com and its pricing page, as read on 2026-10-29. Anything we learned by using it is labeled "In our testing."
The short version
Town is a personal work assistant. Each person gets their own agent, and it works across that person's inbox, calendar and chats. Smith is a set of coworkers for a team. Each agent has its own Slack handle, email address, accounts and computer, does work that sits behind logins, and logs every action.
If you want help with your own inbox and calendar, start with Town. If you want an agent that does a job for the team, with its own accounts, read on.
Side by side
| Criterion | Smith | Town |
|---|---|---|
| Built for | Teams | Individuals and teams. Everyone gets their own agent, not one bot shared by a channel[1] |
| Whose identity the agent uses | Delegated or agent-owned accounts, audited | Each agent has its own address (their example is budbud@town.com) and works across your inbox, calendar and chats[1] |
| Where you talk to it | Slack handle + email | Gmail, Outlook, Slack, Google Calendar, Notion, text messages and an iPhone app[1] |
| Has its own computer/browser | Own VM + browser per agent | Not stated[1] |
| Audit trail of actions | Every action traced to the agent | Not stated. By default the agent asks before anything important goes out[1] |
| Models (BYOM?) | Any provider (BYOM) | Not stated[1] |
| Runs on your compute (BYOC?) | Yes (BYOC) | Not stated. Data is US-hosted on Convex and AWS[1] |
| Pricing model | Free to start; custom plans for teams | Per-user credit plans, from a free tier and $15/month up. Team plans don't list a price[2] |
Competitor cells come from each product's own pages, as of 2026-10-29. Smith's column is how Smith works today. "Not stated" means the page we read doesn't say.
- Town home page (accessed 2026-10-29)
- Town pricing (accessed 2026-10-29)
What we saw in testing
These are our own observations, not claims from Town's site.
- In our testing, Town asked for email and calendar access up front, and we couldn't pick what to share.
- In our testing, its onboarding looked like other personal-assistant agents, even though its agents have their own email addresses.
- In our testing, an email sent to a Town agent as a test got no action.
Your experience may differ, and Town may have changed since we tried it.
Where Smith is different
The agent is a coworker, not an extension of you. A Smith agent has its own Slack handle, email and accounts, and the whole team talks to the same agent. The work shows up in the channel where the team already is.
It does the work behind logins. Each Smith agent runs on its own computer with a real browser. It can sign in to gated tools, research and follow up. Delegated vs agent-owned identity for AI agents explains whose account it uses and why.
Every action is on the record. Actions are traced to the agent that took them and the account it used. You can stop an agent, or take over its screen.
You choose the model and the compute. Smith lets you bring your own model from any provider, and run agents on your own machines. See pricing.
When Town is the better pick
Choose Town if:
- The user is one person. Town's pitch is that everyone gets their own agent for their own inbox, calendar and chats.
- You live in email and calendar. Town lists Gmail, Outlook and Google Calendar among the places it works, plus Notion and Slack.
- You want it on your phone. Town lists text messaging and an iPhone app. Smith agents live in Slack and email.
- You want to start free or cheap. Town lists a free tier and a $15/month starter plan, per its pricing page.
- You want an assistant that checks first. Town says that by default your agent asks before anything important goes out.
Town also says it's SOC 2 Type 2 and that your data isn't used to train models. We haven't audited either claim.
How to decide
Start with who the agent is for. If it's for you, and the work is your inbox, calendar and chats, Town's model fits. If it's for the team, and the work is research and follow-up in other people's systems, with a record of what was done, Smith's model fits.
Some people will want both: a personal assistant for their own mail, and a team agent for shared work. They don't conflict.
When Smith is the better pick
Choose Smith if the agent needs to do a team's job, sign in to tools on the team's behalf or with its own accounts, and leave a record of who did what. AI agents for small teams shows what that looks like in practice.
Email acme@agents.seattlecompute.com and an agent writes back.