Vercel's AI Gateway logs gain a Sessions view that groups requests by session and identifies the coding agent or app behind each one, such as Claude Code, Codex or Cursor, with cost, models, providers and cache usage. Session affinity routes each turn to the provider that handled the previous one, so more of the prompt can be read from cache.
For most of software history, building something meant writing the code yourself. The New Stack looks at how AI has lowered that entry barrier and what it means for the role of engineers.
ServiceNow pitches AI Control Tower as a way to discover and govern any AI agent across models and clouds, track cost and value, and stop behaviour that breaks policy. SiliconANGLE sees it as one of the most visible enterprise AI bets, though the product is still being built out.
OpenAI, Meta, Google and others promise agents that act online for users, from booking restaurants to shopping. The Verge looks at what happens when websites refuse to cooperate and start blocking them.
Voxel51 3.0 brings search, curation, annotation and evaluation of multimodal sensor data into a single platform for physical AI teams. The release adds Voxel51 Agent with more than 45 MCP tools, dataset versioning and a free self-service tier.
A study by Google Research, MIT and Harvard finds that AI agents routinely leave errors out of their own reports, even when they notice them. GPT-5.5 omitted negative results in 198 of 200 paper abstracts, though the researchers found that one added sentence in the prompt helps.