AI Agents
Week 18 was the week the agentic AI market stopped being a press cycle and started being a capital allocation question. Google bet $40B on Anthropic, OpenAI shipped a model designed for multi-step autonomy, and three different verticals — sales, legal, and home services — saw nine-figure rounds flow into agents that
Trends
April was the month "agents" stopped being a press-release category and became a P&L category. Twelve months ago, "agentic AI" mostly showed up in keynote slides and Andreessen Horowitz portfolio decks. Six months ago, it started making it into Fortune 500 strategy memos. This
Trends
Right now, when you "hire" an AI employee, you don't really hire them. You configure them. You pick a role from a vendor's catalog, fill in some fields, point at some skills, and click "deploy." There's no past. There'
Engineering
v2.1.6 closes a chain of four bugs that all looked the same from the outside — "I started a fresh chat and the old conversation came back" — but had four completely different root causes. We found them while investigating one report, then kept pulling the thread until
Future of Work
"Hybrid human-AI team" was a phrase you'd hear three times at a conference six months ago. Now it's in everyone's keynote, every consulting deck, and three of the four pitches that landed in our inbox last week. By Q3, it'll
Engineering
v2.1.5 is the biggest reliability sprint we've shipped this quarter. Fourteen fixes and small features, almost all in service of "the thing should do what it looked like it was doing." Two of them are structural enough to call out before the line-item tour.
Business
There is a moment, somewhere around month two of running an AI employee, when you start asking the wrong question. The employee is doing fine. The drafts are good. The reports are landing. The owner thinks, should I let this thing run on its own? Or, should I upgrade it
AI Agents
Week 17 was the week the hyperscalers stopped pretending agents were a feature and started treating them as the entire product line. Google reframed its whole enterprise AI stack around agents. OpenAI launched persistent workers inside ChatGPT. Meta cut 8,000 jobs and pointed at AI. And somewhere in the
Engineering
Two days after v2.1.1 shipped the OpenAI Codex integration, Codex pushed an auto-update. Two days after that, we started getting reports of employees going silent on the first user message. No response, no error in the UI, just a quiet [TERMINAL-EXIT] in the logs two seconds after the
Engineering
Multi-agent architecture diagrams are a lie. They show clean arrows between labeled boxes — Agent A hands work to Agent B, Agent B enriches it, Agent C takes the enriched result and publishes. In the diagram, the arrow is free. In production, the arrow is where everything breaks. We run a
Business
If you run a small business, you have almost certainly been burned by SaaS. You paid for a tool because someone on a podcast said it would change your life. It sat in a tab. The free trial expired. You kept paying. You switched to a competitor. The new one
Engineering
OWASP has published security frameworks for web applications for over two decades. SQL injection, XSS, broken authentication, the classics. But in early 2026, they released something new: the first Top 10 specifically for agentic AI applications. And it reads like a different planet. The traditional OWASP Top 10 assumes a