What people say
Verdicts and arguments about this category, taken from public posts and quoted one line at a time with a link to the whole thing. A company talking about itself is labelled as that and is never quoted here, only summarised. None of it moves a position on any board: the boards are computed from recorded fields and nothing on this page is one.
About the category
Where the argument is about all of these products at once rather than any one of them.
“if this works meta's behavioral graph becomes a personal context graph. that's a pretty insane extension of an already powerful data flywheel.”
The argument is that ad targeting has always been inference: Meta guessed who you are from behaviour. An agent you hand your email, calendar and finance connectors to does not have to guess, because you tell it, and the same company holds both.
Why it is notable
This is the clearest statement of what an assistant is worth to a company that sells attention, and it reframes every privacy row in our tables: the question stops being whether a product trains on your conversations and becomes what else the parent company already knows about you.
Shopify's president said on launch day that merchants whose products are in the Shopify Catalog and ship to the US are already sellable inside Muse, with no work on the merchant's part, and framed it as the payoff of building on Shopify in an agentic era.
Why it is notable
It records the commercial plumbing arriving before the agents did: an assistant launched in the morning could complete a purchase from an existing catalogue by the afternoon. No listing in this directory can transact in a thread today, and this says the supply side is already wired.
“Companies like Instinct have an enormous ads opportunity. Far greater potential than charging customers directly.”
A founder-investor agreeing with the argument that agents which act on your behalf will make more money brokering offers than charging subscriptions, in a thread calling it the largest user offers exchange yet built.
Why it is notable
Our tables record the published price of every listing and treat a missing price as a disclosure failure. This is the case that the price was never the business model, which would make the pricing column a description of the least important half of the arrangement.
“Concerns from Instinct over conflicts have knocked Index Ventures out of leading a new round for fellow AI assistant startup Town.”
Upstarts Media reported on 3 September 2026, citing three sources, that Index Ventures walked away from a signed term sheet to lead Town's round after Instinct objected, and that Conviction, an investor in both, also stepped back. Forerunner and Menlo took over the round, which Town has not announced.
Why it is notable
Two listings in this directory are close enough that their investors cannot hold both, which is a harder statement about how substitutable these products are than anything either company says on its own site. It is also the first time the competitive shape of this category has been described by someone with no product to sell.
About one product
“A few things I like about Muse vs. single-threaded agents on iMessage or WhatsApp”
An investor's read on what an app buys you that a thread does not: a library of connectors set up in the app, which she expects to be more reliable than driving a browser, and goals that hang off that data with progress attached to them. The screenshots show a connector list running from Bluetooth and Calendly through the Google suite, and a goals screen tracking a vehicle registration renewal and a step count.
Why it is notable
It is the clearest statement of the case against this entire directory. Everything catalogued here reaches you in a thread, and a thread has no room for a settings screen, a connector library or a progress view. If those turn out to be what makes an agent useful, the channel that makes these products distinctive is also what limits them.
“I still feel like the UI is a bit cluttered + there's a few too many things to do + places to click”
The same investor, on the other side of the ledger: too much to do and too many places to click, and no appetite for a feed inside a personal agent. She still expects it to be one of the first consumer agents to reach mainstream adoption.
Why it is notable
The criticism is a channel argument in disguise. A messaging agent cannot have a cluttered interface because it has no interface, which is the strongest thing this category has going for it and the reason the surface area an app buys comes at a cost.
“ChatGPT can do the research too, but I have to come back to it. Instinct brings the answer to iMessage, where my attention already is.”
A user's account of two jobs Instinct took over: widening which retailers he buys from by researching asynchronously and coming back to him, and watching delivery email so it can text him when a package lands. In the exchange he posted, the agent reported the delivery and then volunteered that it was confirmed on paper rather than verified on the ground, and said what would actually settle it.
Why it is notable
Two things a product page will not tell you. The first is the argument for the whole category stated by someone with no stake in it: the research is not the scarce part, the returning to it is. The second is an agent volunteering the limit of its own evidence, unprompted, which is the behaviour that decides whether a proactive message is worth having.
“it has all of the functionality of the imessage agents in flight today but also has all of the ingredients that will expose it to hundreds of millions”
An early user of Meta's Muse says it does what the current crop of texting agents does, with distribution none of them can reach: Instagram's friend graph and the install base of Meta's apps. Posted hours after Muse launched.
Why it is notable
Every listing in this directory competes for the same job, and this is the first public account of the largest possible entrant doing that job with a distribution advantage measured in hundreds of millions of people. Nothing on any product page in the set addresses it.
“Sure Hermes can also do this, but if Muse is doing it it's got integrated access and no risk of being banned.”
A user probing what Muse will actually do on Facebook Marketplace: watch a listing or a saved search for price drops on a schedule rather than instantly, search by query, price, location and condition, and message sellers to negotiate once the Messenger connector is turned on. His conclusion is not about the features. It is that a third-party agent doing the same work is one enforcement decision away from losing access.
Why it is notable
This directory records what an agent can do and says nothing about whether it is allowed to do it, and this is the first plain statement that the second question decides the first. Every agent here that operates a service it does not own carries that risk. The one built by the company that owns the service does not, which is a durable advantage no feature list will show.