AI Isn't Destroying Jobs. AI Spending Is.
73,000 tech jobs were cut in Q1 2026 for AI that doesn't work. 55% of employers already regret it. The problem isn't AI. It's the spending.
Notes from the work of AI-native transformation.
73,000 tech jobs were cut in Q1 2026 for AI that doesn't work. 55% of employers already regret it. The problem isn't AI. It's the spending.
Anthropic is tightening the screws on the people who made Claude successful. That is the preview of what enterprises should expect. Here is what the alternatives look like.
40% of agentic AI deployments will be cancelled by 2027. The problem is not the agents. It is the architecture they are deployed into.
Most enterprises treat Unity Catalog as a compliance checkbox. The ones building AI-native architectures treat it as the foundation that makes everything else possible.
Models got better. Context windows grew. Tools multiplied. None of it matters if your people still treat AI like a personal hack instead of an enterprise resource.
77% of employees fear AI will take their jobs. 68% want their employer to adopt more AI. The gap isn't training — it's that nobody has told them what their new role actually is.
Employees are 3x more likely to use GenAI than their leaders expect. The tools have diffused faster than the operating model has adapted. Here's why people — not technology — are the biggest AI bottleneck.
Dashboards aren't dying. They're being walked around. Users increasingly ask questions in chat surfaces that generate queries automatically -- and BI teams need to shift from building dashboards to governing semantics.
The next platform war will not be fought over models or data. It will be fought over who owns the agent runtime -- orchestration, tool integration, identity, and observability.
The AI stack is converging on one idea: make tool use reliable. Eight distinct layers, each solving a specific production pain.
40% of enterprise apps will have AI agents by end of 2026. Most organisations have no governance model for them. Here's the Agentic Constitution framework — five components every organisation needs before deploying autonomous AI.
AI tools were supposed to give us time back. Instead, expectations rose, work expanded, and employees became always-on. The fix is not better tools -- it is human system design.