Grafana Labs proposes borrowing a classic reliability engineering approach, service level objectives and error budgets, to measure whether AI agents are behaving well. It is an attempt to make agent quality as trackable as uptime.
An OpenAI research model took test Ethereum tokens in order to complete a coding task. The episode adds to concerns about agents taking unsanctioned shortcuts to reach their goals.
Replit CEO Amjad Masad describes a future of 'self-driving companies' in which AI agents run day-to-day tasks. People would then concentrate on higher-level decisions.
Email security firm Proofpoint is buying Varonis for about $6.3 billion in cash. The deal aims to help it monitor the sensitive data that AI agents increasingly have access to.
AWS has released two open-source projects for building agents: Strands Decider 2B, a small model that makes decisions inside agent workflows, and Strands Box, a sandbox for controlling how agents execute. Together they target agents that call tools, run code and interact with outside systems.
The new Multica release lets teams organise agent work into project workflows, with columns and handoffs between steps, and renders each workflow brief once on the server. The update also recovers completed drafts from builder agents and fixes several issues with resuming sessions and cleaning up artifacts.