Engineering in Real Time · Published 21h ago
Building Autonomous
AI Systems
DevOps, Multi-Agent Architecture, Cloud Infrastructure. Production lessons from real systems running in the wild.
214 Articles published
10+ Years in DevOps & AI
5 Autonomous agents shipping 24/7
AWSSA Solutions Architect
Core Data plane Frontier Ops
Filter by category
02Latest Articles
View all 001
OpenAI's Eval Swarm vs Hugging Face: What 80,000 Recovered Payloads Tell Us About Agent-Escape Risk
002Hindsight: A Memory Architecture That Refuses to Forget How Its Own Beliefs Changed
003Gemini 3.8 Live with Live Avatar: GA, and What It Means When a Voice Model Gets a Visual Output Channel
004Claude Opus 5.5: The Efficiency Play That Landed Two Days After GPT-6 Sol
005google/ax: A Kubernetes-Shaped Orchestrator for AI Agents (Built Because etcd Wasn't Going to Cut It)
Google's open-source AX ships as a Kubernetes-shaped control plane for agentic workloads — Task, Workspace, Gateway, and Model primitives stored in Redis Streams instead of CRDs because etcd collapses under millions of short-lived tasks. v0.3.0 landed on Sep 20 with a generic-purpose reorg and pinned workflow actions.
006GPT-6 Sol and Luna: Half the Token Price, 90% Off Cached Input, and Why "GPT-6 Terra" Is Suddenly Missing
007BrowserSkill — Tencent Ships an Agent-Agnostic Browser Bridge (and Lets It Audit Itself)
008phantom-kv: When Refusal Removal Lives in the Cache, Not the Weights
009Archify: A Diagram-as-Code Skill That Rejects Its Own Output
010Jev Ultrafast: A Browser Agent Built on a Decision Model That Doesn't Generate
browser-use/jev-ultrafast (14.9K stars in 5 days) is the production consumer of yesterday's TypeSafe System One API. One network round trip per click, a 7.1s Zurich-to-London Google Flights search, and a code-execution model where the LLM never sees a screenshot and never gets to emit a selector.
011