Town

A per-user agent that joins your group text and talks to other people's agents

Town is an AI agent you reach on iMessage and SMS. To start, create an account. It messages you first rather than waiting to be asked.

iMessageSMS

What does Town cost?: Free to start, with paid tiers. below
Does Town train on my messages?: Qualified statement. the page
Is Town safe?: Holds a scoped token for connected accounts; a security contact is published. the page
Who is behind Town?: Town.com, Inc.; funded. below
What does Town do with my messages?: Who carries them is not published. the page
Town alternatives: 8 that do a similar job. the page

Posted doing a job: 1; public posts, each below
Of those, went wrong: 0; fewer is better; people post successes
Posts for it, against it: 0 / 0; nothing posted yet

How Town works

Town gives everybody a Townie, a named assistant with its own email address that drafts the morning briefing, works the todo list, writes the client proposal and now builds the deck. Its own framing is a colleague rather than a tool, and the pricing follows: free for individuals, then per user for teams.

It publishes more of the things this directory looks for than most listings twice its size. A dated privacy policy, a separate security annex referenced by its data processing addendum, four public price tiers, and a pricing FAQ that explains proration. Very little here has all four.

Two things are worth reading closely. The training answer is a carve-out rather than a position: it says Google user data is not used to train machine learning models, is not used for advertising and is not shared, and it says nothing equivalent about the rest of what a Townie reads. And its own policy states that its third-party advertising cookies and pixels may constitute a sale of personal information or a sharing for cross-context behavioural advertising under California law, with an opt-out honoured for at least twelve months. It is one of only three listings here that disclose an advertising relationship at all.

What Town has shipped

4 features the company has announced, newest first, each linked to the announcement. The examples of what to use it for are the company’s.

2026-09-17
X Daily Digest
A skill you install from a section called Powers: give it the accounts to track and it pulls their posts, ranks them and sends a digest each morning.
Use it for: reading the accounts you follow without opening X; seeding your own posts from what the people you track said.
2026-09-15
Sign-in through 1Password
Announced for the end of September: you approve specific logins in the 1Password app, the Townie signs in through it, and Town says it never sees the credentials or the page while it fills.
Use it for: letting the agent into an account without handing it the password.
2026-09-10
Deck building
The Townie finds the numbers in the CRM, the inbox and the analytics it already sits in, lays them out on brand and hands back a deck ready to present.
Use it for: a board update from the figures the agent already has; a pitch deck laid out on your template.
2026-09-03
Townie-to-Townie coordination
When the answer is in a teammate's inbox, your Townie asks their Townie for it, inside the group thread where everyone can read what was said.
Use it for: settling a time or a detail between colleagues without either of you relaying.

What Town costs, and how you start

Channels	iMessageSMS
How you start	Create an account
Who starts it	Messages you first
Is there an app	No app Nothing to install at any point
Categories	Group agents
Can do	Group threads,Talks to other people's agents,Bookings and reservations,Payments,Web research,Calendar and scheduling,Email,Documents
Connects to	None named publicly
Pricing	Free forever at 30 chat messages a month with every integration included, then Starter at $15 a month, Pro at $49 per user billed annually, and Power at $99. Credits are included with each tier and overage is pay as you go. Enterprise is an annual commitment.
Number type	Not stated on its own pages
Contact number	None published
Documentation	town.com/docs
First message, pre-written	None published
Runs on	Not disclosed
Browses with	Runs its own States it built this itself
Number carrier	No number to look up
Status	Live

Does Town train on your conversations?

Trains on your conversations	Qualified statement“we do NOT use Google user data to train machine learning models”
Data of people who never signed up	Not addressed
What it holds when you connect an account	A scoped token OAuth only, revocable from the provider's side“Third-party service accounts may be connected to the Service via authentication protocols such as OAuth or similar mechanisms provided by your email service provider.”
If you cancel	Not stated in the policy
Model providers named	None named
Jurisdiction	United States. California and other state rights are set out, and the policy states its own advertising cookies and pixels may constitute a sale or a sharing under those laws.
Security document	Published. A security annex referenced from its data processing addendum, last updated June 2026: a written information security programme, encryption for data over public networks, access control, change management, incident management, audit logging.
Asks before acting	Nothing published
Prompt injection	Not mentioned anywhere public
Bug bounty	None published
Availability	Not stated

Whose hands your messages pass throughWho carries them, who runs the model, who drives the browser, and which of those Town names. With what its policy leaves unanswered, counted against the rest of the directory.

Town’s own pages answer 6 of the 9 questions this site asks of every listing. Not answered: says what happens if you cancel, names who carries the messages, names its model providers. About the middle of its kind: 8th of 13 group agents.

Quoted verbatim from the published policy on 2026-09-10, and what a document says is not the same as its legal effect.

Who makes Town

Company	Town.com, Inc.
Founders	Jean-Denis Greze, Tony Vincent
Based in	Not published
Funding	$18M seed led by First Round in March 2025, then a $55M Series A in June 2026 led by Andreessen Horowitz with Forerunner, Alt Capital and Conviction. Reported on 3 Sep 2026 to be raising $90M at a $1B valuation co-led by Forerunner and Menlo Ventures, which Town has not announced. Source: upstartsmedia.com
Investors	Andreessen Horowitz, Forerunner Ventures, First Round Capital, Alt Capital, Conviction
Traction	Not published

From Town’s own pages at town.com, last read 2026-09-10.

An interview

Where the people building this have talked about it on the record. Linked rather than embedded.

1h 16mInterview · 20VC with Harry Stebbings · 2026-09-07Town vs Instinct vs GrokBot | Why the AI Assistant Market Is Not a Bubble
Jean-Denis Grèze with Harry Stebbings. Town's co-founder and chief executive, in a 76-minute conversation recorded as the assistant market was being called a bubble. The published description says Town is reportedly in talks to raise at a $1BN valuation, and the episode is framed around how it differs from Instinct and Grok Bot.
It is the only long-form interview on record with anybody running one of these products, and the three it compares are the three people are actually choosing between this week. The same founder's argument about why Instinct is not a competitor, that one charges teams and the other is fully subsidised, is quoted on the market wall from a post; this is where it comes from.

What people have used Town for

Group coordinationsarah

Group coordination1

2026-09-07sarah 😾 A user added Town to her iMessage group chats, where it stays quiet until mentioned, and then gave it access to the same email address her other agent already used. Two agents from different companies now share one inbox, and she is waiting to see how they behave when they meet.
sarah 😾
@chowtato · 2026-09-07
Having a lot of fun adding @TownAI to group chats. It responds pretty quickly and is holding its own with my friends trying to turn it against me. Next steps is have each friend add their own townie into group chats and make them coordinate our schedules together. Does anyone know why it defaulted to Chinese? Feeling like Kimi might also be the underlying model here. I also added town to the assistant email Instinct uses and now they have shared access. Will be interesting to see how the agents interact with each other.
TownGroup coordination23 likes · 4.3k viewsRead it
The product page describes an agent in your group thread. It does not describe what happens when a second agent, built by someone else, can read the same mailbox: no company controls that surface, neither policy covers it, and the user assembled it herself in an afternoon. The same post also records the agent replying in Chinese mid-sentence and calling it a glitch, which is the kind of thing only a real transcript shows.

In its own words

What Town and its founders have posted: launches, features and claims, recorded as what the product says about itself.

2026-09-17Jean-Denis Greze 💡 Town's co-founder describes a Town skill called X Daily Digest: you give it the accounts to track, it pulls their posts, ranks them and sends a scannable digest each morning, which he uses to seed his own posts. It is installable by anyone from a section of the product he calls Powers, under Explore.
Jean-Denis Greze 💡
@jgreze · 2026-09-17
I don’t really want to scroll through X every morning, so Town does it for me. We have a skill called X Daily Digest. You give it the accounts you want to track, it pulls the posts, ranks them and sends you a scannable digest. So basically every morning I get a summary of what’s happening from the people I care about on X. Then I use that to kick off ideas for things I could tweet as a follow up. This one is something anyone can install. It’s under Powers → Explore in Town.
TownFounder44 likes · 7.2k viewsRead it
A standing job rather than an errand, and a description of how Town's skills are distributed: a catalogue inside the product that any user can install from, which is the shape Muse and Pally describe as connectors and recipes. It is the founder describing his own product, so it is summarised and linked; what it adds to the listing is the existence of a skills catalogue, which the rows here do not yet record for anybody.
2026-09-15Town Town's account lists what it says it is built around: you approve anything before your Townie makes a change, you choose which tools it can use and how, you can review every action, and training is off by default. Then the announcement: a 1Password integration at the end of September, where you approve specific logins in the 1Password app, the Townie signs in through it, and Town says it never sees the credentials or the page while it fills.
Town
@TownAI · 2026-09-15
We build Town around trust and safety. You approve anything before your Townie makes a change. You choose which tools it can use and how. You can review every action it takes. We keep training off by default. Coming at the end of this month: @1Password Approve specific logins in the 1Password app, sign in through Townie, and we never see your credentials or the page while it fills.
TownThe company44 likes · 9.2k viewsRead it
It is the tight direction on the question the password piece asks, the opposite of Instinct enrolling as an authenticator: the secret stays in a vault the person already holds and the agent signs in without reading it. Nothing moves on the listing until it ships, because a row here comes from a page and not a post; when it does, the credential row would go from a scoped token to a vault, the first here that is a third party's rather than the product's own. The training line is worth reading beside the policy: the post says off by default, and the policy this site read says only that Google user data is not used, so the row stays qualified until the policy says what the post says.
2026-09-10Town Town adds deck building. The company's argument is that its agents already sit inside the CRM, the inbox and the analytics, so they can find the numbers, lay them out on brand and hand back something ready to present, instead of the user pulling data across tabs and formatting it.
Town
@TownAI · 2026-09-10
Your Townie can now build your decks. No more pulling data from multiple tabs or wasted hours spent formatting. They already live in your CRM, your email, and your analytics, so they find the numbers, lay them out on brand, and hand it back ready to present.
0:38 video
TownThe company34 likes · 13k viewsRead it
It is the clearest example so far of the work-agent thesis this directory keeps recording from the other side: the value claimed is not the writing, it is the connections that were already there. It also moves Town further from the assistant shape and towards a colleague shape, which is the distinction the categories here are built on, and it fills the documents capability on a listing that had it empty.