Imagine every pixel on your screen, streamed live directly from a model. No HTML, no layout engine, no code. Just exactly what you want to see.
@eddiejiao_obj, @drewocarr and I built a prototype to see how this could actually work, and set out to make it real. We're calling it Flipbook. (1/5)
Generect just made top 3 at Glovo Startup Lab 4.0 π‘
Walking away with $150K in Azure credits and a lot of respect for the Ukrainian founders who shared the stage today.
πΊπ¦πͺ
@The_Cyber_News the per-domain split is the right call. one big 'security agent' tries to do everything and finishes nothing β narrow scope is what lets a subagent actually close a task
Tonight, live, 2 hours: building a native iPhone wellness app from zero with Claude Design/Code and Openclaw
7pm CEST (in an hour) https://t.co/WeyJdDCVdm
ran a board meeting today where 12 AI employees reported to 1 human
0/3 commitments executed for the 7th week in a row
the agents are honest about it though. every metric, every miss, no spin.
turns out the hard part of AI-first teams isn't the AI. it's the human in the loop.
MCP is getting the same disease APIs had
people think standard transport = standard behavior
it doesn't
if auth, errors, side effects, and object names all differ, the agent still needs custom glue for every tool
protocol solved the socket
product still has to solve the contract
@polsia facts. most CRM AI is still a polite intern sitting next to the database
real unlock is when the agent can do the boring ops work directly and safely
read, update, move stage, log why
chat on top is the garnish