Signal history
Every week that made it into the loop.
Browse every published edition with its weekly context, publish date and a direct link to the reader.
-
Week 30 July 20-24, 2026013
The week agent security became real, production controls hardened and software defense sped up
This week made one thing clearer: AI is moving from demos into real systems, and the harder questions now are about containment, permissions, production controls and software defense at machine speed.
Published
Read edition
Friday, July 24, 2026 -
Week 29 July 13-17, 2026012
The week of open models, tighter guardrails and infrastructure that can no longer stay invisible
This week made one thing clearer: AI, software delivery and financial infrastructure are all getting more automated, which makes control surfaces, trust boundaries and operational guardrails much more visible.
Published
Read edition
Friday, July 17, 2026 -
Week 28 July 6-10, 2026011
The week of stronger agents, weaker benchmarks and harder software guardrails
This week sharpened three realities at once: frontier models are becoming more usable, coding benchmarks are less reliable than they look and the security model around agents is getting much more urgent.
Published
Read edition
Friday, July 10, 2026 -
Week 27 June 29-July 3, 2026010
The week of agent governance, model access and controlled frontier releases
The week leaves a clearer picture of where AI is landing in real work: stronger agents, tighter safety controls, harder debates over who gets frontier access and more structured releases for powerful models.
Published
Read edition
Friday, July 3, 2026 -
Week 26 June 22-26, 2026009
The week of Vercel, Jalapeño and AI controls
Vercel exposes the risk of delegated AI access, OpenAI moves deeper into custom compute with Jalapeño, and the week closes with stronger controls across software, infrastructure and banking.
Published
Read edition
Friday, June 26, 2026 -
Week 25 June 15-19, 2026008
The week of GitHub Actions, OpenAI Enterprise and SpaceX-Cursor
GitHub hardens CI/CD, OpenAI moves enterprise AI toward more control, SpaceX buys Cursor, and digital fraud forces stronger defenses.
Published
Read edition
Friday, June 19, 2026 -
Week 24 June 8–12, 2026007
The week of SpaceX's IPO, Claude Fable and resilient identity
SpaceX's IPO, Claude Fable, security in the tooling and more automated defense.
Published
Read edition
Friday, June 12, 2026 -
WWDC 2026 June 8–12, 2026006
WWDC 2026 Special: Apple tries to return to the center of personal AI
The signals that actually matter from WWDC 2026: Siri AI, Gemini under the Apple Intelligence architecture, rebuilt search, performance, automation and new layers of family safety.
Published
Read edition
Monday, June 8, 2026 -
Week 23 May 29 – June 5, 2026005
The week AI arrived through AWS, Vite and the budget
A curated read of the week in infrastructure, development, security and digital fraud.
Published
Read edition
Friday, June 5, 2026 -
Microsoft Build 2026 June 2–3, 2026004
Top 10 Special Edition: Microsoft Build 2026
Ten signals from Microsoft Build 2026: agents in production, in-house models, enterprise knowledge, local AI and data for agentic applications.
Published
Read edition
Tuesday, June 2, 2026 -
Week 22 May 25–29, 2026003
The week of Claude, npm and product security
AI, development, security and product. A short read to decide what to look at more closely.
Published
Read edition
Friday, May 29, 2026 -
Week 21 May 15–22, 2026002
The week AI started becoming work infrastructure
AI, development, design, security and digital trust. A short read on how agents go from answering prompts to coordinating workflows.
Published
Read edition
Friday, May 22, 2026 -
Week 20 May 10–15, 2026001
Agents Everywhere: from isolated tools to connected agents
AI, design, development and security enter a new cycle: fewer isolated tools and more agents connected to the real workflow.
Published
Read edition
Friday, May 15, 2026