From the archive
Friday, 20 November
Folded from 34 sources that day.
Sorted by quality and by how many sources covered it, the same for everyone.
-
Fragments: November 20
An experiment on 465 open-source pull requests found unhealthy code raised the cost of agent review by 22 to 35%. The roundup also covers AI for reading terms of service and slowing down AI software factories.
-
Your AI agent's "memory problem" may be something else
Agent failures often mix three kinds of context: workflow state, durable memory and external knowledge. Each needs its own owner, lifecycle and access rules, even when a checkpointer stores them together.
-
Presentation: Modernising Retail at Scale: Architecting a Cloud-to-Edge Platform for a Global Enterprise
Jayesh Bhayani describes Tesco's decade-long move from fragmented legacy systems to a unified cloud-edge platform. He shares lessons on build vs. buy, cloud cost control and preparing for agentic commerce.