Watching live tv and working with Claude works well
Match up promoting with ads so you’re giving full attention.
This is good for night work
During the day, you should be working across 4-5 veins (proper labeling - I use numbering + task name - helps with context switching).
Just do what works.
Start with the complex, learn the language of systems and practice solving problems.
Then just keep it simple. Do what works and do it well. Follow the playbook, improve iteratively, and test everything.
Pure transformer/harnesses at I/O layer can’t judge consequences of actions taken.
You’re gaining alpha off of historicals from a paradigm that you’re disruption.
Short term wins are accessible with capital and tooling. Long term alpha comes from within/networks.
Plus reliability is an open. Agents display extreme intelligence when they maintain fidelity over 100,000 steps+ as context grows without degradation. But propping up global financial markets on top of agents that rely on guardrails is just a house of cards waiting for an edge case #NassimNicholasTaleb
Yep. Innovators dilemma is one thing. Trying to rip out the fully system (human workflow, data governance, enterprise architecture) mid flight with reporting requirements? Good luck.
Velocity of growth from ground up builders beats time to adopt…and that’s if you’re measuring by a 30+ year vet exec trying to learn how to properly leverage cowork much less run a system of agents.
At some point nueroplasticity comes into play as well
Yeah we’re pretty much in this boat. It’s a tricky one to avoid, at least for our focus, due to trade offs.
I’ve built for markets that needed the product but lacked the tech fluency before. It’s a loosing battle. The adoption curve on agents is no joke and really makes it hard to leave the bay (especially as someone that was first brought here by those same market forces a few months ago).
The hard truth is that most of the conversation around AI/agent adoption is performative and the highest concentration of real action (with capital behind them) is the technical leaders of Bay Area series A+.
This is for us, which is a bit of a custom fit to the issue given our positioning. But I’d say the above is the key driver to a broader trend.
Ya know beyond readily available funding (and customer connects from that funding), social proximity, and general lack of creativity/individual thinking from this generation of founders
@tailopez It still isn’t. Agents only perform well on tasks <10 steps. Avg human task is 50+ steps (in terms of complexity).
Current level of agents handle tasks, they need to handle objectives before they can truly serve as employees
steve jobs didn't drink athletic greens. plato didn’t take creatine. picasso didn’t wear a whoop. shakespeare didn’t own an eight sleep. einstein didn’t use an infrared sauna. they all did just fine.
Unless someone is clearly building/contributing to the distributed creation of this next chapter for humanity…don’t waste your personal time talking about AI.
This is the most interesting time in human existence. If that’s not enough to spark your curiosity, then please remove yourself from the conversation
parents: "move out"
girlfriend: “quit being such a loser”
boss: "work harder"
claude: "uber for dogs (the dogs are the drivers) is a great idea, you should absolutely pursue it"