AI agents are getting good at tasks. The limiting factor is trust.
You'll give an agent your calendar before your email. Email before your CRM. CRM before your bank.
The whole agentic future runs on a permission curve nobody's mapped yet.
Visa is embedded in ChatGPT. Mastercard launched Agent Pay for Machines.
AI agents can now shop, pay, and transact autonomously.
The last bottleneck was money and now that's being solved for.
Last night, between rounds of comments from his MD, my friend and I got into it over dinner: which industry builds AGI first, Tech or Finance?
The hedge fund argument is compelling. Skin in the game and ruthless cost discipline.
But that's exactly why I think labs get there first.
"Best output at lowest cost" is an efficiency function. AGI is a capability problem, not an efficiency problem.
So these are my reasons why the labs win:
The final mile requires burning hundreds of billions on things that mostly won't work - a fund can't justify that to LPs. OpenAI lost money for years and that's basically the playbook.
Then there's talent. People who understand scaling laws, RLHF, constitutional AI - they're at Anthropic and OpenAI, not at Citadel.
And Renaissance's Medallion Fund is the sharpest counterpoint to the whole fund thesis. Over 30 years they've had the best quants in the world, the strongest possible motivation (returns). They're the peak version of the hedge fund argument.
But they didn't build AGI, they built a very good pattern -matching machine for financial markets which is a fundamentally different thing.
So: a fund won't build AGI, but it'll be the first to know AGI exists, through an impossible P&L.
Which brings me to the more interesting question: when AGI lands, who owns the interface layer?
Labs build the model and financial institutions help detect it. But who builds the harness that you and I actually use?
Investors are already positioning for the outsized returns these products will generate, we saw it on YCs request for companies this batch.
Companies like @joinsauna are already well on their way to being in exactly the right place when AGI arrives.
Those are the companies I want to be early on.
Vibe coding let developers build apps without touching code.
Vibe doing is what comes next. You don't build a thing that does the work. You just tell the AI... do the work.
That unlocks the other 950 million knowledge workers who never cared about software.
Cursor closed the gap between what you're thinking and what's in the file.
Vibe doing is that bet for everyone else.
@kozerafilip Have been waiting for this to drop for so long. I was on the beta version and have just look at what the GA looks like, freaking insane. Will promptly cancel my Claude lol
Proactive agent that thinks and acts like you.
Multiplayer AI Brain for teams.
Proper GUI for commanding 50 agents.
https://t.co/zsDxICHezv is all three. Sauna goes live today. First 2000 people, use access code LAUNCH for $80 of weekly(!) credits.
Let’s explain. Multiplayer only works once the personal brain is powerful. So let's start here.
Personal AI Brain
3,800+ tools connected. State of the Art memory. Skills and schedules you teach once that get repeated forever on cheaper models. An AI first CRM.
Lives on the cloud so you can initiate tasks from anywhere: iMessage, Slack, Email. Also no need for a Mac mini 😉
GUI
AI agents have been stuck in their MS-DOS era. A chat box, a scroll buffer, no way to command 50 of them. We built the first GUI: Live sessions on one side, work waiting for your sign-off on the other, plus the things Sauna kicked off while you were asleep waiting for review. Game mode helps clear the queue with actual joy.
Okay so far so good, but how to give benefit of what you built to more people or whole team?
Multiplayer
Once your Sauna actually knows you and you gave her access to your tools, you can use multiplayer.
Two modes:
- Brain access. My co-founder Robert plugged my brain as a tool into his Sauna last month. He can ask it about pricing while I'm in other meetings, gets a sourced answer back, never has to interrupt me. That’s read only. Yolo mode gives him access to all my tools too :O
- Communal Saunas extend that to whole companies with proper permissioning. Folder owners decide what's true for the whole company and build skills. Most get their personal brain + read access to communal files and memories. Works also for group planning my best friend's bachelor party.
The Way
I’ve been obsessed about AI Brain since 2016. Our human brains suck at some things like memory and are brilliant at others like creativity. We are also particularly bad at thinking we are all on the same page and then realising weeks later that we weren’t. Most leaders spend their days being the human diff tool, catching contradictions in hallway conversations and Slack threads. Repeating themselves 50 times. Now every company is spinning up hundreds or thousands of agents that drift faster than humans do, feeding each other their drift as context. Compounding rot.
After 10 years and one failed company in this space we are launching the solution.
The Launch
We thought about a celebrity launch. Margot Robbie in a bathtub explaining agents. But then we realised the same money gives the first 2,000 people free daily credits, every day, until we burn through that $1,000,000.
Sauna runs Claude, GPT-5, Gemini, GLM-5.1, DeepSeek, Kimi so you can budget yourself. The labs are racing to lock you so they can milk you in a year. We picked your side.
Onboarding
It’s live at https://t.co/8ekc1ZtzoJ We’ve preheated saunas based on a niche. Use the access codes in the comments to get a better experience.
In our beta, the average customer paid us $299 and cost us $1,024 to serve.
That gap is the AI agent market right now. General access opens Wed.
Every AI founder in horizontal products i competing with the big labs. If you are aiming at building a trillion dollar company you join the bloodbath, eat the margin, and bank on models getting cheaper.
We raised 30mm so we can participate in the war. That doesn't make the war sane. The short term winners are the customers but the part they should worry about is what happens after lock-in. Anthropic grew to $30B in ARR in a very short amount of time. That growth has to be sustained somehow. Over time, they will need revenue growth and real margins, especially on the path to becoming a public company, and the cleanest way to get there is to charge locked-in customers more. They already started.
If your company has built all its workflows, agents, automations, and habits around one model provider, every future price increase lands directly inside your operating system.
For a lot of knowledge tasks, the raw intelligence is already good enough. The bottleneck has moved to the agent harness and the product: can it pull the right context, choose the right model, call the right tools, remember what happened, and show the work through surfaces people actually want to use?
That is where we think we're much better. Our bet is better agent harness, model choice, multiplayer and UX.
Sauna gives you the best intelligence available today, including Opus 4.7, and right now we are eating a lot of that cost. But the long-term product is built to get cheaper.
Skills can be tested against specific models. Scheduled tasks can run once on the strongest model, document the work, and then run repeatedly on smaller models that are good enough for the job. The expensive model becomes the teacher. The cheaper model becomes the operator.
The durable value is the layer above the models: context, memory, integrations, permissions, and multiplayer. Shared agents with shared company context are genuinely hard to build well, which is why most AI products are still solo toys with expensive models attached. We think the product layer is where this market gets won.
We launch free access to Sauna on Wednesday. Big things are coming.