@devahaz@TheRideshareGuy He has extremely high conviction on his opinions, his positive bets are funded and his negative bets are rarely funded to the same extent. For instance he didn’t find a way to synthetically short SF over the last 5 years with actual dollars. This works in VC, prob not much else
@icanvardar More likely they just drop a new model and keep costs high my bro. I don’t see how this eventuates without open source or a few more frontiers entrants to keep margins low competition high. Still some job losses but on the margins
@NathanFlurry lab harness and model decides how many agents / tokens to use on specific task. the incentive between the user and the model provider is not there at the moment.
there used to be a cost to niche'ing down which stopped the microsofts of the world going down stack to compete with legal tech / accounting tech / property tech etc. there is no or v little cost now. labs study the use of their models, pick the next biggest use case and embed in the harness as a new tool / use case.
@Jason at this point startups are just outsourced R&D and market validation mechanisms for the labs. i love their products but hate the macro impacts
they're are cottoning on that OAI / Anthropic dominance ruins their ability to continue operating as VCs. Lets just call a spade an anti-competitive monopoly already an apply some level of anti-trust stay-in-your-lane-ism.
OpenAI is trying to own every app and platform and sell tokens, which means they want to kill every startup.
That’s their right, it’s a free market.
we need open source alternatives built in America that are committed to enabling startups with tokens — instead of selling them tokens and using the profits to replace them.
@kentcdodds It’s a good game of political chicken. if Dario isn’t lying about existential job loss then it’s justified, if he is lying then the bubble bursts or at least takes some pressure off accelerationism.
Excited to share our latest HEX work for @composio.
The company recently raised a $25M Series A led by @lightspeedvp and is building the infrastructure layer that helps AI agents interact with software.
Creating a brand for a company that's carving out its own category comes with a unique set of challenges.
The identity is rooted in the idea of electricity. Bright beams, motion, and interconnected systems reflect the infrastructure Composio provides behind the scenes – connecting tools, coordinating actions, and powering the flow of information between agents and software.
We built a visual system designed to feel technical, distinctive, and adaptable; something that could grow alongside the product as AI agents become a bigger part of how software gets built and used.
Proud of where this landed.
Huge thanks to the Composio team for trusting us with the project.
If you're looking for a rebrand ahead of your next fundraise or product launch, I'd love to chat.
if inference speed goes to effectively zero wait time then working with agents will become a much more focus driven experience. right the multiagent orchestrating process is a constant context switching tax while you wait for any agent to complete its task / goal