Big announcement!
I'm now a co-founder at @electricsql along with the @thruflo / @balegas and the brilliant team they've built.
We're solving local caching of Postgres data.
Electric is a VC-backed devtools OSS startup. We're close to our first beta release so check us out!
Your agents don't live in a sandbox, they live in a function and a log.
Split state and compute.
We've built the architecture for the future of agentic software.
The world needs managed agents. The architecture for them is serverless.
Pull apart the harness. Run the agent logic in a function and wire the tool calls into your existing data platform and backend systems.
Every major AI platform has now shipped managed agents. On the same sandbox-based architecture.
That architecture is wrong. Managed agents don't belong in sandboxes. They belong in functions 🧵👇
Every major AI platform has now shipped managed agents. On the same sandbox-based architecture.
That architecture is wrong. Managed agents don't belong in sandboxes. They belong in functions 🧵👇
@joodalooped@NickGideo@ibuildthecloud@schickling@aboodman 👀 curious what's not usable about our stack? We use it a lot & have a ton of customers/users now — always things to improve of course but happy to hear about what rough edges you ran into
@joshuamanns Hey we've had our hands full scaling the system for some big customers. It's not super trivial to solve unfortunately. Doable but will need some dedicated time
CC: You're right—the goal said "do not stop until the entire plan and all phases are finished," which means all 14 phases, not just Phase 3. I stopped after completing Phase 3 because I treated "the plan" as the Phase 3 plan I had loaded, rather than the full demo plan
😬
Alan Kay's "objects as like biological cells and/or individual computers on a network" formulation of programming is gonna finally have its moment with agents.