The bull case for assistant *startups* is that Muse is bootstrapping the 3P connector ecosystem, which can then plug into Instinct / Tomo / Sxn etc.
Unlike iOS, these “apps” are interoperable — just an MCP.
Unlike the OG fb platform, they don’t depend on the graph (yet!)
Unlike iOS, these “apps” are interoperable — just an MCP.
Unlike the OG fb platform, they don’t depend on the graph (yet!)
Sep 23, 2026 · 4:46 PM UTC
21
9
154
28,008
@illscience So now, you think there will be 8 to 9 appstores that matter? I get that MCP is the "protocol", but how would these apps get distribution? I am assuming that there will be a native ad? You just don't want RockYou (and the other startup that @mlevchin built) to be FB'ed.
2
0
1
469
@Simonkhalaf @mlevchin unique vertical to peel off or find a way to outcompete horizontally.
There may be more opportunities than we imagine though. For example, some people prefer a gruff doctor and otherwise prefer a warm and chatty one. A single agent can’t be both, so it’s possible markets like health get segmented by personality and capability.
There may be more opportunities than we imagine though. For example, some people prefer a gruff doctor and otherwise prefer a warm and chatty one. A single agent can’t be both, so it’s possible markets like health get segmented by personality and capability.
1
0
3
311
@illscience The bear case, of course, is that Muse and forthcoming offerings from OAI/Anthropic/Google/Apple/etc. just eat them alive.
The distribution advantage is hard to overcome when the products are more or less the same.
The distribution advantage is hard to overcome when the products are more or less the same.
1
0
6
400
@MatthewSchrager For sure though that’s traditional competition vs being locked out of an app/platform moat ..
0
0
4
211
@illscienceny and VC always focus on revenue (money) and nothing of consumer safety it seems. It’s like in banking where they’ll be willing do wrong for 10!years because the fine is minuscule in comparison to the gains. I may be wrong but which actually presents a more positive return long term or is it only short term that counts? Do you all invest in products that actually have an accountability thesis with a long tail focus in mind?
2
0
2
333
@Maybejoshb I guess the counterfactual is that we have hundreds of millions of consumers that overpay on loans / cards / accounts but do so in a “predictable” way… imo our bar should be better than humans / status quo not perfect
0
0
2
193
@illscience No doubt they will interlink their data advantage from graph to act as a persistent stateful layer for Muse.
Bit like Spotify - they offer a commodity which you can get anywhere, but once seasoned to your taste… good luck getting a consumer to switch
Bit like Spotify - they offer a commodity which you can get anywhere, but once seasoned to your taste… good luck getting a consumer to switch
1
0
1
276
@GlobalTMTMoose Very likely especially on targeting / discovery .. but other agents may use the capabilities in different ways so still potentially accretive
1
0
1
176
@illscience @grok what does graph mean in the above context?
1
0
0
16
@illscience Exactly. It is a behavior and business model change that will be difficult for a startup like us to convince the big players. Muse is going to make the ecosystem more friendly to this.
Then it comes down to who can deliver more value to both sides.
Then it comes down to who can deliver more value to both sides.
0
0
1
172
@illscience muse also takes plain apis and clis, not just mcp, so do those carry over to instinct too?
0
0
1
147
@illscience The "(yet!)" is the part I keep thinking about. Connectors move between assistants, but who my people are and what each assistant can share with them doesn't. Do you think that graph ends up owned by one assistant, or as its own layer?
0
0
0
154
@illscience I made this! Check out my account, Relay can connect to any AI service and assistant and has type safe messaging between agents.
1
0
0
11