mini agents?

I think there’s a much bigger story here than Cursor potentially losing access to OpenAI models What’s happening across AI right now is that the model layer is fragmenting faster than anyone really anticipated, and honestly I think that’s a good thing There won’t be one model that is best at everything forever, there will be different models that are better at different kinds of reasoning, different price points, different latency requirements and different types of work, and the best products will increasingly be the ones that can move between all of them without making the user think about it That makes the relationship between model companies and the applications built on top of them much more complicated though Cursor is a pretty good example because the product itself is not really about OpenAI or Anthropic or any particular model, it’s about giving someone the best possible coding experience and using whatever intelligence is best to make that happen The model is becoming a component of the product rather than the product itself And as that happens, I think the industry is going to need infrastructure that sits above individual model providers and keeps the ecosystem open This is one of the reasons I have a lot of respect for what OpenRouter is doing As the number of models keeps exploding, having a neutral layer through which applications can access and compare intelligence becomes much more important than simply having another API The interesting thing about OpenRouter is that it lets the market decide what is useful A model can come from a company nobody has heard of, suddenly perform incredibly well on real workloads, get picked up by developers and start competing with models that have spent billions building their brand That kind of competition is extremely healthy for AI But it also means applications should not have to rebuild themselves every time the model landscape changes or become dependent on whether one model provider decides they want to allow a particular use case If intelligence is going to become as abundant and competitive as I think it will, then access to that intelligence needs to become more open and composable too This is also a big part of how we think about the GThinking Wand at Supermission I don’t think the future is users opening a product and choosing between twenty model names before they can get something done That’s just pushing infrastructure complexity onto the user If I ask an AI to solve something difficult, I shouldn’t have to know whether the best answer will come from OpenAI, Claude, Gemini, GLM, DeepSeek or some model that gets released six months from now The system should figure that out It should understand the problem, decide how much reasoning is actually needed, choose the right intelligence, bring in another model when it makes sense, challenge its own answer when the stakes are high and keep adapting as the underlying models improve That is the layer we are interested in building Because I think the biggest shift in AI over the next few years is going to be from model-centric products to intelligence-centric products Models will become increasingly interchangeable, while the systems that know when and how to use them will become increasingly valuable So while this looks like a disagreement between OpenAI and Cursor on the surface, I think it is also a signal of where the architecture of the AI industry is heading More models, more fragmentation, more competition and eventually much more abstraction between the person using AI and the model doing the thinking The user shouldn’t have to care which model thought about the problem They should just care that it was the right intelligence for the job

$25B in DEX volume on Robinhood Chain This is what onchain adoption actually looks like: real users, real liquidity, and real trading happening at scale. And as more markets move onchain, the opportunity for autonomous trading gets bigger. That’s the market we’re building for at Supermission. AI agents that can find, evaluate, and execute opportunities across DeFi, prediction markets, and perps. The rails are here, comes the intelligence

gmgm, everyday a new look to experience, check this one out more towards improving the data and signals and correcting incoming streams throughout the site, to the moon $MISSION incoming 👀

3,000,000 $MISSION this one is for the people who actually helped build the mission for the contributors who showed up early for the people who tested, gave feedback, shared ideas, helped others, created, pushed through the rough days, and kept believing when there was nothing to prove yet we’ve always believed that the people building alongside us should have a place in what comes next so we’re putting 3,000,000 $MISSION aside for a community giveaway not for insiders not for a handful of wallets for the contributors who have been part of this journey you helped shape what $MISSION is becoming now it’s time to give something back more details on eligibility and distribution soon this one is for the builders🫡 drop your address 👇

we are on $MISSION impossible it's already live on robinhood chain what happens next this coming month: → trading fees go back to holders → platform revenue buys $MISSION on the open market → $MISSION becomes the currency of the agent economy → shipping log published every month

someone said an app szn?

We are excited for the partnership and great to see infra powering @WizzHQ data core agents The future is agentic, more coming, stay tuned

“ok so why is agentic trading the future” Glad you asked. ▶︎ •၊၊||၊|။||||။၊|။|||။|||။၊| 4:52:29

🚨🚨 FOR SUPERMISSION We’ve been selected for the OpenAI for Startups program We’re building the next generation of AI-native trading -combining AI agents with prediction markets, perps, and real-time market intelligence. Grateful to be building at the intersection of AI × markets × agents. supermission.fun