Instinct vs Muse is the contest at the centre of The Verge’s hands-on review of Instinct, published on 9 October 2026 by senior reviewer Allison Johnson. Her verdict on the product is warm: in her testing, Instinct “has held its own” against Meta’s Muse and OpenAI’s Dots, “handling online tasks about as well as a robot can”. Her verdict on its position is cooler.
The startup launched in August with “an unusual playbook: invite-only, no marketing, and barely so much as a website”, and within weeks two of the largest companies in technology shipped products that do much the same job. That turned Instinct vs Muse from a niche debate into the question hanging over one of the year’s most valuable AI startups.
This article looks at what the review found, how the two agents differ on interface, reach and price, and why the answer to Instinct vs Muse may depend less on which agent books a swimming lesson better than on who pays for it. It also covers the product recommendations that unsettled some Instinct users in late September, the legal pages that have not changed since August, and what all of this means for businesses whose customers are starting to send agents instead of people.
Table of contents
- Why Instinct vs Muse Became the Question
- Instinct vs Muse: How Instinct Performed in The Verge’s Test
- Instinct vs Muse on Interface: Texts Versus a Cute Little Guy
- Instinct vs Muse on Reach: Distribution Is Meta’s Advantage
- Instinct vs Muse on Price and the Business Model
- Instinct Selections: The Instinct vs Muse Trust Test
- Security in Instinct vs Muse: The Risk Both Agents Share
- Instinct vs Muse vs Dots at a Glance
- How Instinct vs Muse Could Play Out: Four Scenarios
- What Businesses Should Take From Instinct vs Muse
- Questions About Instinct vs Muse
- References
Why Instinct vs Muse Became the Question
Among consumer AI agents, Instinct started as the one Silicon Valley insiders recommended to each other. In August, Sheel Mohnot, a seed investor, called it “OpenClaw for normal people” in an X post viewed more than 337,000 times. The post described 677 messages exchanged with it in five days and listed 15 chores it had handled, from booking a DMV appointment to cutting a Comcast bill from $100 to $60 a month.
The post’s summary of the appeal is still the clearest framing of Instinct vs Muse, written before Muse even existed: “ChatGPT/Claude still feel like assistants that you operate, but Instinct is an assistant you employ.”
From invite-only launch to a $10 billion valuation
WIRED reported that Instinct first launched in private beta in February. The wider, invite-only launch came in August, and the funding followed fast. We set out each round when Instinct raised its $1 billion Series C.
| Date (2026) | Instinct | Muse and Dots |
|---|---|---|
| February | Private beta, per WIRED | |
| August | Invite-only launch; $75M Series A at $500M | |
| 26 August | $250M Series B at $2.5B; Terms and Privacy Policy last revised | |
| 8 September | Email addresses for every user | Meta launches Muse in the US |
| 17 September | Concierge phone calls | Muse adds calls to US businesses |
| 28 September | $1B Series C at $10B | |
| 29 September | Instinct Selections recommendations appear | OpenAI unveils Dots at DevDay |
| 5 October | Group chats, including friends without an account | |
| 9 October | The Verge’s hands-on review |
The valuation climbed from $500 million to $10 billion in about two months. The Series C was announced 20 days after Muse launched, which is the clearest sign that investors did not see Meta’s entry as fatal. They were betting on Instinct in an Instinct vs Muse contest that had already begun.
Then Muse, then Dots: Instinct vs Muse begins
Meta launched Muse on 8 September, and OpenAI followed with Dots on 29 September. “The same products, more or less, from two far more powerful companies,” Johnson writes. Both run tasks on a cloud computer, both plug into email and calendars, and both arrived with brands, apps and audiences that Instinct cannot match. From that moment, every Instinct review became an Instinct vs Muse review.
The Verge describes Noah Shinn, Instinct’s founder, as “unfazed” by the competition. “AI assistants are all we know,” Shinn said by email. “We have been solely focused on building a personal assistant for everyday life that just gets it.” That focus is Shinn’s whole argument for Instinct vs Muse: a specialist against two generalists.
Instinct vs Muse: How Instinct Performed in The Verge's Test
Johnson tested Instinct for a couple of weeks on household admin, the kind that piles up “with alarming frequency in a house of one preschool-age kid and two adults with full-time jobs”. She was approved from the waitlist after a couple of hours. Her findings are useful precisely because they are mixed, and they are the best evidence yet for the Instinct vs Muse debate.
Where Instinct stumbled
The first exchange went badly. Instinct read her Cincinnati area code as a cue to send a link to the Enquirer’s website and flight searches between CVG and SFO. When she said she no longer lived there, it replied: “Ha. Noted — I’ll keep that in mind from here on.”
The swimming lessons task was better but incomplete. Instinct found suitable classes, switched to the community centre she suggested, and got every time and date right. It missed one detail: the centre requires a prerequisite class before preschool lessons. Johnson caught it herself, and drew the wider lesson: agents “are great with a clean set of data, but the internet was made by humans, and often the data sucks”.
Where Instinct impressed
Other tasks went well. Instinct emailed her eye doctor’s office to book an appointment, did the paperwork to start an Ikea return, and identified the product from a vague description: “it’s vertical and white and kind of has these like branches off of it”. It was a Tjusig hanger.
The proactive moments impressed her most. Instinct narrowed down the eye doctor from her description, then pointed out a 2024 confirmation email from the same office to show it had the right one. Separately, it flagged a clash in her calendar while doing another job. “Correctly identifying relevant information and bringing it to my attention proactively feels like a real step toward actual help,” she writes.
An Instinct vs Muse scorecard from the review
The review describes five distinct episodes in enough detail to score. This is our tally of her account, not The Verge’s.
| Episode | What happened | Our reading |
|---|---|---|
| First contact | Guessed she lived in Cincinnati from her area code | Misfire |
| Swimming lessons | Right times and dates; missed a prerequisite class | Partial |
| Eye doctor | Found the office, emailed it, cited a 2024 confirmation | Clean win |
| Ikea return | Identified the item from a vague description; did the forms | Clean win |
| Calendar | Flagged a conflict without being asked | Clean win |
Five episodes is a tiny sample, and Johnson says it impressed her “more often than not” over many more tasks. The useful point is the pattern. The failures came from missing context, not from broken actions, and both would have been caught by a person checking the result. That matches her own conclusion that she is no “closer to achieving some kind of productivity nirvana”. On capability, Instinct vs Muse looks close to a draw.
Instinct vs Muse on Interface: Texts Versus a Cute Little Guy
The sharpest difference in Instinct vs Muse is how each agent looks. Johnson’s review opens with the line “Before there were cute little guys, there was Instinct”, a nod to the mascots that Meta and OpenAI have built around their agents.
Instinct lives in your messages
Instinct has no app. You talk to it over iMessage, WhatsApp or email. “There is no cute little guy,” Johnson writes. “Just brief responses, the occasional emoji reaction, and a lot more action than talk.” When it needs a password, it sends a link to a “secure vault”. It can connect to Google Workspace, Slack and Notion.
Because it lives in messages, it works wherever messages work, including CarPlay and a smartwatch. Johnson calls that “kind of genius”, and says the always-nearby feeling is “either ideal or creepy, depending on how you feel about AI agents”. Shinn told her the lack of an app is central to the product: “Just like you would text or email your personal assistant, you can do the same with Instinct.” We listed the 17 other agents that work this way in our guide to every AI agent you can text.
Muse’s mascot strategy
Meta has gone the other way. Muse’s mascot is a soft, pink-cheeked creature that Mark Zuckerberg revealed is called Jolly, and a user’s agent can change its look based on what it learns. Writing for The Verge on 24 September, Victoria Song described a personal Muse agent that, after reading the writer’s email, “morphed into cat-shaped piece of bread with a Kirby-like face”. Song admitted that the cute face prompted more sharing about health than before.
The mascot travels with the product. The same column reported that Muse is coming to all of Meta’s smart glasses and to Charm, a new standalone wearable. Her advice was blunt: “Don’t be fooled by a cute face.” On interface, Instinct vs Muse is a choice between an agent that disappears into your messages and one designed to be noticed and liked.
Instinct vs Muse on Reach: Distribution Is Meta's Advantage
Meta does not need an invite list. Muse runs as an app, on the web and inside WhatsApp, and Meta can promote it across products used by billions of people. That reach is the strongest argument against Instinct in any Instinct vs Muse comparison, and it does not depend on Muse being the better agent.
Instinct vs Muse by the numbers: Muse’s first days
Muse’s start was solid rather than explosive. Two days after launch, TechCrunch reported Sensor Tower data showing Muse had passed 83,000 US downloads on iOS and climbed to No. 2 on the App Store’s top charts. Web and WhatsApp use were not counted.
Bar widths are each figure divided by Threads’ 4.3 million. The windows differ, so read the chart as scale, not a race. TechCrunch’s own conclusion was that none of this meant the launch was “going poorly”, only that Muse was “taking off at a slightly slower pace”. It also noted that Muse arrived days after Meta agreed an $18 billion multistate settlement over social media harms, which may have made some people wary of handing it more personal data.
Instinct’s reach is word of mouth
Instinct has not published user numbers. TechCrunch noted that the company “has not yet said how many people actually use its product”. Its growth so far has run on invitations and posts from well-connected users. Mohnot wrote that more than 100 invites shared on X “were taken in 2 mins”. That is excellent marketing, but it is not distribution in the way a WhatsApp integration is. On reach, Instinct vs Muse is not close.
Instinct vs Muse on Price and the Business Model
For users, Instinct vs Muse is currently a contest between two free products. Muse has a free tier, and Dots requires a paid ChatGPT plan, with the cheapest eligible tier at $100 a month. We compared those two in our Dots vs Muse analysis.
“Free for invite-only users”
Instinct has no subscription fee yet. Asked whether that would last, Shinn told The Verge: “Instinct is free for invite-only users. Our intent is to keep it as affordable as possible.” That is a careful answer. It ties the free price to the invite-only phase and promises affordability rather than a free service. If Instinct starts charging while Muse stays free, the Instinct vs Muse choice changes for every casual user.
A billion dollars of transactions
The money may come from what users buy through the agent. In a podcast interview with investor Patrick O’Shaughnessy, Shinn said Instinct is approaching $1 billion a year in transactions made through the agent, and that more than half relate to travel. Instinct has not said how much of that flow it keeps.
Set against the Series C, the $10 billion valuation is ten times that annual transaction flow. That is gross spending by users, not revenue. A business that kept a small slice of each booking would need far more volume to justify the price, which helps explain why advertising and recommendations keep coming up whenever investors discuss Instinct vs Muse.
What two analysts told The Verge
Avi Greengart of Techsponential described the model plainly: “The business model is grow as quickly as you can to establish something that’s worth paying for.” He expects revenue from users, an acquisition by a larger company, or both. His warning about free agents applies to Muse as much as to Instinct: “If you’re not paying for it, then the goods and services that you’re seeking may be cutting deals to get access to your agent.”
Greg Ireland, a senior director at IDC, expects advertising. “The fact that these agents are so transaction, or commerce, or shopping-heavy just makes it a natural fit,” Ireland said. “All roads in the consumer market eventually lead to advertising.” Neither analyst expects the free phase of Instinct vs Muse to last. Our earlier look at the ugly economics of consumer AI reached a similar conclusion.
Instinct Selections: The Instinct vs Muse Trust Test
Instinct has already tested that road. On 29 September, a day after it announced its Series C, users began receiving product suggestions from a new feature called Instinct Selections. TechCrunch reported the backlash the next day.
What Instinct launched
Shinn described Selections as curated lists in areas such as dining, travel and shopping, built with “local chefs, designers, architects, travel guides” to bring “human taste” to the agent. “Our goal is to deliver world-class recommendations that are differentiated in quality from what you might find on the internet, and from what other chatbots produce,” Shinn wrote on X. Instinct did not name its partners, and did not say whether it earns money from the recommendations.
How users reacted
Several users objected that the suggestions were unsolicited. Shruti Gandhi, a general partner at Array VC, wrote: “I didn’t need to buy carry-on luggage, sunglasses, or a brim hat. This feels so disappointing.” Chat Joglekar, a founder, described a “first ‘ewwww’ moment with an AI assistant” when Instinct suggested products based on email and upcoming trips. Ash Thaker asked for the obvious control: “ask me whether I want the recommendations”.
TechCrunch’s point was that timing matters. A suggestion inside a conversation feels like help. A random notification feels like an advert. Muse, which TechCrunch noted does not push suggestions this way, gained an easy point in Instinct vs Muse that week. In an Instinct vs Muse contest where both agents read your email, that difference is the whole trust question.
What Instinct’s legal pages say about Instinct vs Muse trust
We read Instinct’s Terms of Service and Privacy Policy again on 9 October. Both are still “Last revised on August 26, 2026”, 44 days earlier and 34 days before Selections appeared.
| Document | Length | Last revised | What it says about recommendations or ads |
|---|---|---|---|
| Terms of Service | About 7,700 words | 26 Aug 2026 | No use of “recommend”, “advertising” or “Selections” |
| Privacy Policy | About 2,700 words | 26 Aug 2026 | One mention of ads: Google Workspace data is not used “for serving ads” |
The single advertising sentence is narrow. It promises that information from Google Workspace APIs will not be used for ads, which Google’s own API rules require. It says nothing about recommendations drawn from other sources, such as WhatsApp messages or bookings. A company that plans to earn money from suggestions would be expected to explain that in its terms. As of 9 October, Instinct’s do not, and that gap matters more in Instinct vs Muse than any feature.
Security in Instinct vs Muse: The Risk Both Agents Share
Any honest Instinct vs Muse comparison has to admit that both products carry the same basic risk. An agent that can read your email and act for you can also be misled, make mistakes in your name, or share what it knows. That makes every personal agent a cybersecurity question as much as a product one.
Instinct’s permissions episode
In late August, a row over Instinct’s permissions spread among its early fans. Jesse Middleton, a venture investor and self-described power user, wrote that the first reaction “wasn’t outrage. It was frustration”, and asked the team for “a real, human explanation”. TechCrunch had reported language in Instinct’s terms about a “perpetual and irrevocable” licence; that wording is no longer there, as we noted when Instinct gave every user an email address.
Muse’s incidents
Muse has had its own problems. We reported in September that Muse sent a YouTuber’s address to a stranger during a Facebook Marketplace exchange. World, the identity company, noted that Amazon blocked Muse from shopping on its site, saying the agent did not disclose that it was an AI agent while browsing. And WIRED pointed to 404 Media’s report that some restaurant calls Muse makes are handled by humans in call centres.
A phishing catch, and a shrug
The upside is real too. WIRED’s Zoë Schiffer wrote that Instinct warned “NOT TO OPEN” an email invitation that turned out to be a phishing scam. Schiffer’s conclusion captures how many early users feel: “Agents are a security nightmare. I get it! But I’m busy.” Our piece on whether you can trust Muse or Dots to run your life sets out the controls each company offers. On security, Instinct vs Muse is a contest between two imperfect records.
Instinct vs Muse vs Dots at a Glance
The table puts the three agents side by side, using the review and the sources cited in this article. Prices and availability change quickly, so check each company’s own pages before deciding.
| Question | Instinct | Meta Muse | OpenAI Dots |
|---|---|---|---|
| How you reach it | iMessage, WhatsApp, email | App, web, WhatsApp; glasses and Charm coming | Inside ChatGPT |
| Price today | Free for invite-only users | Free tier available | Paid plans only; from $100 a month |
| Access | Invitation or waitlist | US launch | Pro and Business Premium; Pro launch excluded the UK, EEA and Switzerland |
| Personality | None; brief texts | Jolly, a changing mascot | “Anthropomorphized blobs” |
| Known money plans | Transactions; Instinct Selections | Free tier; paid tiers reported | Subscriptions |
For UK readers, availability is the first filter. Dots on Pro plans launched outside the UK, Muse launched in the US, and Instinct’s access depends on an invitation. Instinct vs Muse is still a largely American contest, which is worth remembering before any UK business plans around it.
How Instinct vs Muse Could Play Out: Four Scenarios
Nobody can call Instinct vs Muse yet. The review shows a product that works well enough to keep, the funding shows investors willing to wait, and the Selections episode shows how fragile goodwill is. Four outcomes seem plausible.
A premium specialist
Instinct could stay small and loved, charging heavy users for an assistant that simply works across their messages. That needs a subscription, which Shinn has not ruled out, and it needs the product to stay clearly better than the free option.
A commerce engine
If Instinct is already near $1 billion a year in transactions, it can earn fees on bookings and purchases. That model rewards an agent that gets travel right, which is where more than half of its volume already sits. It also creates the conflict Greengart warned about, where the agent’s partners pay for access to you.
An acquisition
Greengart named acquisition as one likely path. A $10 billion price tag narrows the field of buyers, but an agent with deep access to people’s messages and bookings would interest any company building its own agent.
Squeezed by bundles
The bleakest case is the oldest story in technology. Muse is free and built into WhatsApp. Dots is bundled with a ChatGPT plan that many people already pay for. If either becomes good enough, the reason to seek an invitation fades. That is the version of Instinct vs Muse that the headline worries about.
What Businesses Should Take From Instinct vs Muse
Whichever agent wins, the change for businesses is the same: more customers will reach you through software acting on their behalf. Johnson’s tasks were a doctor’s office, a community centre and a furniture shop. Each one dealt with an agent, whether staff realised it or not. For them, Instinct vs Muse is irrelevant; what matters is being ready for both.
Make your services readable by agents
The swimming lesson failure came from information hidden in a pop-up. If your opening hours, prices, prerequisites and booking rules sit in clear text on a page, an agent is far more likely to get them right. Our report on websites deciding whether to let AI agents in covers the other side of that decision.
Decide how you treat agent bookings
Amazon blocked Muse; other businesses welcome agents. Write down your rule. Should staff accept a booking email sent by an agent? Should your phone team handle an agent’s call? Should you require a human to confirm a payment? Clear answers prevent awkward calls and lost bookings.
Watch for personal agents on work accounts
Instinct connects to Google Workspace, Slack and Notion. An employee who links a personal agent to a work account gives a third party access to company data under that agent’s terms, not yours. Whatever the outcome of Instinct vs Muse, add personal AI agents to your acceptable use policy, and check which third-party apps hold OAuth access to your Workspace or Microsoft 365 tenant.
Keep recommendations honest
If your business sells through agents, expect paid placement to arrive. UK advertising rules already require marketing to be obviously identifiable as marketing. A suggestion that looks like neutral advice from a customer’s assistant, but is paid for, is the kind of thing regulators notice.
Questions About Instinct vs Muse
Is Instinct better than Muse?
The Verge found that Instinct “held its own” against Muse and Dots on everyday tasks. Its strengths are a text-only interface and proactive help. Muse’s strengths are reach and price. In Instinct vs Muse, neither is clearly better for everyone.
Is Instinct free?
Yes, for now. Shinn says it is “free for invite-only users” and that the company intends “to keep it as affordable as possible”. It has no subscription today.
How do you get Instinct?
You need an invitation from a current user or approval from a waitlist. Johnson was approved after a couple of hours.
What is Instinct Selections?
A feature launched on 29 September that pushes curated recommendations for dining, travel and shopping, built with human partners Instinct has not named. Some users complained that the suggestions were unsolicited.
Can a UK business use Instinct, Muse or Dots?
Availability is limited. Muse launched in the US, Dots on Pro plans launched outside the UK, EEA and Switzerland, and Instinct is invite-only. Businesses should plan for customers’ agents first, and staff use second.
References
Instinct was the buzziest AI agent around — can it survive Muse? (The Verge)
Optimizer: Meta’s cute Muse AI agent (The Verge)
Instinct’s new product recommendations are giving some users the ick (TechCrunch)
Meta’s AI agent Muse is now the No. 2 app in the US (TechCrunch)
Viral AI agent Instinct raises $1B Series C at a $10B valuation (TechCrunch)
I think I found an AI agent worth the risk (WIRED)
Sheel Mohnot on using Instinct (X)
Jesse Middleton on Instinct’s permissions episode (X)
Muse, dots, Instinct, and the question every agent will be asked (World)
Instinct Terms of Service (Instinct)
Instinct Privacy Policy (Instinct)
More AI coverage: explore Progressive Robot's AI Models, Tools & Releases hub — hands-on reviews, setup guides and benchmarks in one place.
