Instinct AI vs Town AI: price, invite, and who asks first
One agent reads the tool manual before acting. The other works from a list narrowed in advance.
Read the
manual.
Observed reachable commands require their help text to be loaded before execution.
571 actions in its help index
See a
shortlist.
The catalogue is narrowed before the main assistant sees it.
177 shown from 1105 catalogued tools
The counts describe different surfaces, so they do not measure capability or accuracy. Read how we counted them.
Three differences beyond tools
What it is
Instinct Lives in your messaging apps rather than in a new interface.
Town Web and native Mac, with the broadest native reach into your Mac of the four.
Where work runs
Instinct uses a leased cloud browser for browser tasks. Its desktop build was switched off on our account.
Town has a native Mac app that reaches messages and the screen.
When you approve
Instinct shows a card at the moment of acting. You approve the exact text on it.
Town sets approval in the routine's configuration in advance, with extra approval possible on individual tools.
What you can change
Instinct uses three memory stores; one is designed not to return what it holds.
Town has a personality document you can open and edit.
What the same tasks showed
Two kinds of evidence follow: what we observed in the products, and what happened on shared tasks using adapted runtimes. The task results do not tell us which product decision caused an outcome.
Separate batches of adapted runtimes using a substituted model, not end-to-end results for the consumer apps. No combined score or ranking.
October
20 shared tasks, one attempt each
Instinct40% (20–64) · 15 tasks
Town15% (5–36) · 20 tasks
September, same tasks
114 shared tasks, older rules and builds
Instinct52% (45–61) · 114 tasks
Town37% (30–44) · 114 tasks
September, own draw
Each product answered its own draw, so rates are not directly comparable.
Instinct43% (36–50) · 159 tasks
TownFigure withheld
95% interval measured rateOctober's intervals overlap; September's underlying transcripts have not been republished. The batches cannot be compared as a trend. Product runtimes rebuilt for the exam with one substituted model, not the consumer apps measured end to end. Results and caveats
Instinct AI vs Town AI splits on access, price and when you are asked. TechCrunch reported that Instinct launched invite-only in August 2026, and in The Verge's hands-on access meant an invitation from a current user or approval from a waitlist. Instinct publishes no price. Town publishes its prices, starting with a free plan. Instinct asks on a card at the moment of acting. Town asks earlier, in the settings of a routine.
They also disagree about tools. Instinct's help index lists 571 actions; we recovered full manuals for 335, and did not verify every action's reachability. Observed reachable commands require their help text to be loaded before execution. Town catalogues 1105 tools and then hides five sixths of them, showing its main assistant 177. Memory, execution and channels are separate observed differences, not demonstrated consequences of that choice.
- Access differs before any task starts: Instinct's examined route was invite-only with no published price; Town offered web signup and a published free plan.
- Both can start from a trigger rather than a message you type. That makes execution place and approval timing practical questions.
- The two common-task batches below are separate measurements of adapted runtimes; neither is a verdict on the shipping apps.
Instinct is made by Spear Street Technology, Inc., which its terms of service name as the operator. What its terms say about paid services is on our Instinct page. Town is made by Town.com, Inc. You sign up on the web, there is a native Mac app, and the tiers are on our Town page. Town's own site also lists an iPhone app, Slack, texting from any phone, WhatsApp and Telegram, and says the three chat channels are for asking, since Town cannot reliably message you first there; we have not examined those. For how this pair sits among the others we opened, see Instinct vs Muse and where personal agents run.
Fortune’s report describes Town’s $55 million Series A led by a16z, its late-2024 founding by Plaid’s former CTO and its personalised assistants, called Townies.
Instinct vs Town tools: a manual to read or a short list
Instinct- Help index571 actions
- Read the relevant manual
- Execute the command
Town- Catalogue1105 tools
- Narrow in four stages
- Main assistant sees177 tools
These are different kinds of counts. The diagram shows where each product narrows the agent's view, not which can do more.
Instinct does not ship a function list. It ships a command-line surface: 43 families and 571 actions by the product's own count, and three different totals depending on who is counting. An action refuses to run until its own help text has been read, which makes the documentation part of the prompt and therefore part of the bill.
Town ships a catalogue of 1105 tools narrowed in four stages before anything is offered. The main assistant sees 177. One background writer is allowed 17.
These are different tool-loading mechanisms; we did not measure their effect on choice accuracy. Instinct's answer is to make the agent pay attention: observed reachable commands require loading their descriptions first; not every registry entry was verified. Town's answer is to make the choice earlier, on the agent's behalf, and in a place the agent never sees.
There is a second-order difference too. Instinct's rule puts the documentation inside the loop, which means the product's own help text is load-bearing: if a description is wrong or stale, the agent could be wrong in a way that looks like a model failure. Town's narrowing happens in code the agent never reads, so a bad narrowing rule could produce an agent that reports it cannot do something it can. We did not test either failure; they are different enough that the same error would be looked for in two different places.
Why the Instinct and Town tool counts cannot be compared
A comparison article would put 571 against 1105 and conclude something. Neither number means what it looks like.
Instinct's three totals use different counting methods: 571 actions in its own help index, 569 in our pass over that index, and 577 distinct action names once live pages are added, of which 335 provided a full manual. Town's 1105 is a catalogue rather than an offer, and the number that governs behaviour is 177, or 17 if the work falls to that background writer.
The honest comparison is not a count at all. It is that one product's limit is enforced at read time and the other's is enforced at selection time, and only one of those is visible to the agent doing the work.
Tool counts are easy to quote and say little on their own. A tool count tells you how much the vendor has built. It does not tell you how much is reachable in a given conversation, which is the only version of the question that affects what happens to your errand.
A leased browser against a Mac that can see your screen
Instinct orchestrates on a server with the work running in a sandbox, and browser tasks go to a leased cloud browser rather than the Chrome on your desk. On the account we examined we saw no work run on your own machine: Instinct has a desktop build, but a feature flag kept it shut, so we never saw it run. The leased browser can use sign-ins you have saved.
Town handles sync and sessions server-side but ships a native Mac application that genuinely reaches into the system: messages, contacts, screen and audio.
This is the same axis that separates Pine from Town, and the conclusion is the same: the exposure differs more than the capability does. What makes the pairing with Instinct interesting is that it inverts the tool story. The product that lists the most actions to its agent kept the examined work in a sandbox and a leased browser. The product that shows its main assistant a sixth of its catalogue has a native foothold on your Mac.
Town and Instinct on memory: one you can edit, and a vault that never gives anything back
Town keeps a wiki, a memory store and a people model on the server, and the agent's personality is a document you can open and edit rather than a line buried in a prompt.
Instinct keeps three stores doing three different jobs: a git-tracked set of markdown memory files, an indexed layer of connector data it can query, and a fill-only vault for secrets. That last one is the detail worth sitting with. It is built so that credentials can go in and never come out, which is a deliberate and good design, and it also means there is a part of what the product holds about you that neither you nor the agent can inspect.
The difference in practice is who edits. Town gives you a personality document to open and change. Instinct keeps what it knows as markdown files under version control, so a corrected fact stays in the history rather than vanishing, and the product decides what goes into them. We did not establish whether you can open those files yourself.
Neither arrangement is careless. A vault that cannot be read back is the right design for credentials and a worse design for everything else, and the question is which things ended up inside it. An editable personality document is the right design for a disposition and a riskier one for anything a confused user might overwrite. What a comparison can say is that Town is the one where we confirmed a document you can go and change when it has the wrong idea.
Where you reach Instinct and Town, and who else can start them
Instinct is messaging first. WhatsApp, SMS and iMessage are the default surface, plus an address on a mail domain it gives you. The web app is mostly an admin panel. It can also start itself: a clock, or an event on your mail, calendar, documents or location, with a platform-side check before a wake becomes an action. And it can now pick up the phone: TechCrunch reported a concierge feature that calls businesses for you, for things like appointments at places without online booking. We have not examined that feature.
Town gives web and Mac equal weight, and routines can be started by a clock, an email or a calendar entry rather than by you typing.
Put those beside the execution paragraph and the asymmetry sharpens. Both can be woken by something on your calendar. Town can then act through an application that reads your messages. Instinct supports the examined messaging channels and sandboxed work; connected external accounts can also be affected. Neither is the safer product in general. One concentrates risk in what it can touch, the other in how easily it can be addressed.
Instinct vs Town: which one asks before it acts
TownSet run mode, account scope, tools and approval mode in the routine.
InstinctRead the card and approve the exact text it shows.
Town may still ask again for an individual tool. Instinct's card covers actions, not reading.
Both can start without you, so this matters more here than in most pairings. They answer it at opposite moments.
Instinct asks at the moment of acting. In the account we examined, an approval card carries the plan, names the device that will run it, and offers approve, deny or dismiss. Approval binds to the exact text shown, and a rewrite needs a fresh approval. Its own rules forbid an unseen assumption from riding along with an irreversible send, and it will give a receipt on request listing what it read, what it sent and what it did not touch. The card covers actions, not reading: in one reported case it found a login code in a connected inbox and used it without asking.
Town asks in advance, through configuration. The run mode, the account scope, the tool list and the approval mode together decide whether a task may start at all and whether it may write to anything outside, and an individual tool can carry extra approval on top. Approval records are kept alongside its model of the people the assistants deal with. Town's control row has the detail.
The practical check follows. With Instinct, read each card, because the exact text is what you are approving. With Town, read the routine's settings before it is ever triggered, because a routine started by an incoming email runs under whatever you configured when you set it up, plus any extra approval an individual tool carries.
The same gap in the Instinct and Town teardowns
Our page for each product carries a list of what we could not establish, and the overlap is almost exact.
For Instinct we could not get the body of the main instruction set, only its section headings, nor which host the main agent runs on, nor the production model id, nor four named policy sections we found pointers to but no text for. For Town we could not get the full instruction set, because the member-facing interface refuses it, nor where the personality document is spliced in, nor the execution bodies of its routines.
Both products let us see the shape of the instruction set and not its contents. That is not coincidence, it is the boundary of what external examination reaches, and it is worth stating in a comparison because it is the part where any writeup, including ours, stops knowing and starts inferring. The two lists above mark where our confidence ends.
Instinct vs Town: which one the choice comes down to
Two questions decide it, and neither is answered by a feature list.
Does the work depend on things the agent can only learn by looking at your computer? If yes, Town's examined Mac app reaches that context and Instinct's examined routes did not; its desktop build stayed shut on our account. If no, Town's foothold is exposure carried for nothing.
Do you want to reach the agent from wherever you already are? Instinct is built on that premise, and the design follows from it: no new app to open, no window to check, everything surviving as a text message. Town, as we examined it, expects you to come to it, on the web or on the Mac, and in exchange it is present on the machine in a way a message thread can never be.
Put as fit rather than as a ranking:
Pick Instinct if you want to hand over errands from WhatsApp, SMS or iMessage, you would rather approve actions on a card as they come up, and the work does not depend on what is on your computer.
Pick Town if the work depends on your Mac's messages, contacts or screen, and you are willing to decide in advance, in its settings, what it may do unattended.
What the figures do and do not settle
Our September scorecard carries a completion rate for Instinct with its interval and task count; the table near the top of this page prints it. Town's September figure is withheld, because the figure we were given does not reconcile with the only records we can check.
That withheld row is part of the honest state of this pairing. We can compare these two products on how they are built, which is what this page does. Two separate measurements below do give both products the same tasks: an older September batch, with its filtering and provenance limits stated, and the October batch. Why we print a rate only with its interval and sample size, and never a combined score, is set out in how scoring works.
On the separate 114-task batch, measured on 22 September 2026, Instinct completed 52% (95% interval 45 to 61%; 380 attempts) and Town completed 37% (95% interval 30 to 44%; 513 attempts). Both answered the same 114 tasks. These are replicas of older runtimes with the same GLM 5.3 flash model substituted, under older rules, filtered from roughly 5,600 attempts using eight rules, not fresh product results. Transcripts were supplied but have not been republished here; some files were overwritten or shared between attempts, so the underlying record is incomplete. Keep this measurement separate from the main table and do not treat interval overlap or a point estimate as a significance test.
The October batch, run on 11 October 2026, gave every product the same 20 tasks once each, under the exam rules that took effect at the end of September. Instinct completed 6 of 15 scored attempts, 40% (95% interval 20 to 64%), and Town 3 of 20, 15% (5 to 36%). With one attempt per task, one result more or less moves a rate by five to seven points, and the two intervals overlap. The tasks, the rules and our replicas all changed since September, so neither October figure says a product got better or worse, and the batches are never added together. The other October batches, each with its own tasks and builds, are listed separately on the results page, and individual recorded attempts are on the run pages, as counts rather than rates.
For price context, the Town snapshot includes a free plan, Starter at $15/month, and Pro at $49/month billed annually, checked 2026-11-05. Other monthly paid prices retain their earlier snapshot date, as the pricing note explains.
Full teardown matrix
The rows below preserve the exact scorecard wording for each product, including what we could not establish. They are here for reference; the comparison and its evidence limits are explained above.
Show the scorecard rows
| Axis | ||
|---|---|---|
| Where it runs | Server orchestration with work running in a sandbox. Browser tasks use a leased cloud browser, not the Chrome on your desk. | Server-side sync and session handling. The Mac app is native and genuinely reaches into the system: messages, contacts, screen and audio. |
| How tools are exposed | A command-line surface rather than a function list: 43 families and 571 actions by the product's own count, and three different totals depending on who is counting. An action refuses to run until its own help text has been read, which makes the manual part of the prompt. | A catalogue of 1105, narrowed in four stages before anything is offered. The main assistant sees 177 of them; one background writer is allowed 17. |
| Where memory lives | Three stores that do different jobs: a git-tracked set of markdown memory files, an indexed layer of connector data it can query, and a fill-only vault for secrets. | Server-side: a wiki, a memory store, a library of documents and routine output, and a people model. The personality is a document you can open and edit, not a line buried in the prompt. |
| What you get to approve | An approval card with three answers rather than two: approve, deny, or dismiss. It carries the plan and names which device will run it. Approval binds to the exact text shown, not to the request that produced it, so a rewrite needs a new approval. Its own rules forbid an unseen assumption from riding along with an irreversible send, and it will produce a receipt on request: what it read, what it sent, what it did not touch. Deletion has a documented surface covering storage, training, location and desktop indexing. | Approval is configuration rather than a prompt. The run mode, the account scope, the tool list and the approval mode together decide whether a task may start at all and whether it may write to anything outside, and an individual tool can carry extra approval on top of that. Approval records are kept alongside the people the assistants deal with. Memory has its own add, update and delete tools rather than a setting. |
| How you reach it | Messaging first. WhatsApp, SMS and iMessage are the default surface, plus an address on its own mail domain. The web app is mostly an admin panel. | Web and Mac carry equal weight, and routines can be started by a clock, an email or a calendar entry rather than by you typing. Town's own site also lists an iPhone app, Slack, texting from any phone, WhatsApp and Telegram, and its help pages say the three chat channels are for asking, since Town cannot reliably message you first there; we did not examine those. |
| What we could not establish | The body of the main instruction set (only its section headings surfaced), which host the main agent runs on, the production model id, and four named policy sections we found pointers to but never text for. | The full instruction set (the member-facing interface refuses it), where the personality document is spliced in, and the execution bodies of its routines. |
| Completed, scored subset | 43%36 to 50 · 159 tasks | Withheld |
- Where it runs
- InstinctServer orchestration with work running in a sandbox. Browser tasks use a leased cloud browser, not the Chrome on your desk.
- TownServer-side sync and session handling. The Mac app is native and genuinely reaches into the system: messages, contacts, screen and audio.
- How tools are exposed
- InstinctA command-line surface rather than a function list: 43 families and 571 actions by the product's own count, and three different totals depending on who is counting. An action refuses to run until its own help text has been read, which makes the manual part of the prompt.
- TownA catalogue of 1105, narrowed in four stages before anything is offered. The main assistant sees 177 of them; one background writer is allowed 17.
- Where memory lives
- InstinctThree stores that do different jobs: a git-tracked set of markdown memory files, an indexed layer of connector data it can query, and a fill-only vault for secrets.
- TownServer-side: a wiki, a memory store, a library of documents and routine output, and a people model. The personality is a document you can open and edit, not a line buried in the prompt.
- What you get to approve
- InstinctAn approval card with three answers rather than two: approve, deny, or dismiss. It carries the plan and names which device will run it. Approval binds to the exact text shown, not to the request that produced it, so a rewrite needs a new approval. Its own rules forbid an unseen assumption from riding along with an irreversible send, and it will produce a receipt on request: what it read, what it sent, what it did not touch. Deletion has a documented surface covering storage, training, location and desktop indexing.
- TownApproval is configuration rather than a prompt. The run mode, the account scope, the tool list and the approval mode together decide whether a task may start at all and whether it may write to anything outside, and an individual tool can carry extra approval on top of that. Approval records are kept alongside the people the assistants deal with. Memory has its own add, update and delete tools rather than a setting.
- How you reach it
- InstinctMessaging first. WhatsApp, SMS and iMessage are the default surface, plus an address on its own mail domain. The web app is mostly an admin panel.
- TownWeb and Mac carry equal weight, and routines can be started by a clock, an email or a calendar entry rather than by you typing. Town's own site also lists an iPhone app, Slack, texting from any phone, WhatsApp and Telegram, and its help pages say the three chat channels are for asking, since Town cannot reliably message you first there; we did not examine those.
- What we could not establish
- InstinctThe body of the main instruction set (only its section headings surfaced), which host the main agent runs on, the production model id, and four named policy sections we found pointers to but never text for.
- TownThe full instruction set (the member-facing interface refuses it), where the personality document is spliced in, and the execution bodies of its routines.
- Completed, scored subset
- Instinct43% · 95% interval 36–50% · 159 scored tasks
- TownWithheld
FAQ
What is the main difference between Instinct and Town? Access, price and when you are asked. Instinct launched invite-only and publishes no price. Town publishes prices, starting with a free plan. Instinct asks on a card at the moment of acting. Town asks earlier, in a routine's settings. They also load tools differently: Instinct exposes 571 actions and requires the help text to be read first, and Town catalogues 1105 tools and shows its main assistant 177.
Which has access to more of my computer? Town, as far as we examined. Its Mac application is native and reaches messages, contacts, screen and audio. Instinct ran work in a sandbox and sent browser tasks to a leased cloud browser; its desktop build was switched off on our account, so we never saw it run.
Can I change what either one thinks about me? Town's personality is a document you can open and edit. Instinct keeps a git-tracked set of markdown memory files, an indexed layer of connector data it can query, and a vault built to accept secrets and never return them. We did not establish whether you can edit Instinct's memory files yourself.
How do I reach each one? Instinct lives in WhatsApp, SMS and iMessage, plus a mail address it provides; its web app is mostly an admin panel, and a clock or an event on your mail, calendar, documents or location can also start it. Town treats web and Mac equally and can also be started by a clock, an email or a calendar entry. Town's site also lists an iPhone app and messaging channels we have not examined.
How do Instinct and Town do on tests? We do not rank them. Instinct has a September completion rate on our scorecard with an interval and a task count; Town's September main-table figure is withheld because it does not reconcile with the records we can check. In the separate older 114-task batch, Instinct is at 52% (95% interval 45 to 61%) and Town at 37% (30 to 44%). In the October batch of 20 shared tasks, Instinct is at 40% (20 to 64%) and Town at 15% (5 to 36%). The batches are never combined, and the caveats above apply.
Who makes Instinct and Town, and what do they cost? Instinct is made by Spear Street Technology, launched invite-only in August 2026, according to TechCrunch, and publishes no price. Town is made by Town.com, Inc. and publishes its prices, including a free plan. Both pricing pictures are on the teardown pages, Instinct and Town, so they stay in one place when they change.
Is this based on testing or on their documentation? On our own teardowns of both, Instinct and Town, including the lists of what we could not establish for each.
Other comparisons
- Muse vs InstinctCompared on what each one actually runsInstinct AI vs Muse, from our teardowns of both: where each runs, how you reach it, which asks before acting, phone calls, and which fits your errand.
- Pine vs TownA borrowed browser or a Mac footholdPine AI vs Town AI: Pine thinks on its servers and borrows a browser; Town's Mac app reaches your messages, contacts, screen and audio. From our teardowns.
- Muse vs PineAnd what a machine of your own is actually forMeta Muse vs Pine AI: Muse gives every user a Linux VM in Meta's cloud; Pine runs its model on its servers and borrows a browser. From our own teardowns.
- Muse vs TownAnd two things people call a machineMeta Muse vs Town AI: Muse gives each user a Linux VM in Meta's cloud; Town's native Mac app reaches your messages and screen. From our own teardowns.
- Pine vs InstinctTwo products that look identical from one step backPine AI vs Instinct AI: both think on a server. Pine is built around calls and borrows your browser; Instinct lives in your messages and leases one.
All six pairs, and why there are only six, are on the comparisons page.
More on Instinct
- Instinct teardown
- How Instinct AI worksHow does Instinct AI work? One errand traced from trigger to reply: what wakes it, how it picks tools, where it runs, what it remembers and when it asks.
- Instinct AI reviewInstinct AI review roundup: what reporters who used it got done and got wrong, how that maps onto our teardown, and our exam figures with intervals.
- Is Instinct safe?Is Instinct AI safe? Every incident users and reporters have described, from unchecked emails to login codes read from Gmail, checked against our teardown.