Feed wire.
Latest dispatches on the wire — plus a short, dated view of the work in motion.
Now
Celln
Building an execution plane for agents: isolated cells that borrow verified tools rather than assembling mutable environments. The immediate work is making the real KVM path easy to install, inspect and prove.
Sympozium
Exploring a clean split between coordination and execution. Sympozium should own workflow, policy and shared state; Celln may become the smaller execution primitive beneath an AgentRun.
Questions I am following
- What does a useful persistent agent API require beyond one-shot jobs?
- Which capabilities should an agent request explicitly: tools, files, egress, time, memory?
- How do we make revocation and provenance operational rather than aspirational?
Updated August 2026 · read the Celln notes →
Latest dispatches
5 items · also available as an RSS feed
Testing Celln's Boundary
Four boundary probes and one successful computation: what a small Celln demonstration shows, and what it leaves untested.
The Agent Environment Should Be a Lease
Why I built Celln to lend agents a defined capability set, and what that execution model still needs to prove.
Post-Kubernetes Infrastructure for GenAI Workloads
Field notes on Modal's million-sandbox announcement, what it says about Kubernetes' assumptions, and the coming decoupling of coordination from execution.
The Sticky-Note Problem: Making Agent Handoffs Reliable
Why agent handoffs lose context, and how shared evidence and explicit ownership can make coordination more reliable.
Agents Need Somewhere to Share Their Work
What makes agents useful as a team, and how a shared, policy-controlled workspace might help.