TikTok Launches Ads MCP Server for AI Agents. TikTok opens an MCP endpoint that lets external AI agents plan, launch, and optimize ad campaigns without manual intervention. This turns ad platforms into agentable endpoints and forces outcome engineers to design for permissioning, audit trails, and on-platform agent flows — Principle 09.

Fetch.ai launches AEVS for verifiable agent executions. Fetch.ai ships AEVS to publish tamper‑evident, on‑chain cryptographic receipts for each agent action. This gives a practical path to verifiable agent executions and post‑hoc audits, shifting how you build Gate and Validation controls — Principle 16.

Notion courts developers with a platform for AI agents and workflow automation. Notion launches a developer platform and hosted Workers to embed AI agents, sync enterprise data, and automate workspace workflows. This shows major collaboration platforms are becoming agent runtimes; outcome engineers must own connectors, context plumbing, and lifecycle controls for enterprise data — Principles 06 & 09.

Microsoft’s multi-agent AI system tops Anthropic’s Mythos on cybersecurity benchmark. Microsoft’s MDASH coordinates 100+ specialized agents across models to outperform Mythos on CyberGym and speed vulnerability discovery. This demonstrates that scaled specialization and agent coordination drive capability, so build orchestration patterns, role decomposition, and robust validation pipelines into your stack — Principle 09.

Zerops raises $2M seed to rebuild cloud infrastructure for the AI development era. Zerops raises funding to unify dev and prod on a bare‑metal PaaS that lets AI agents build and deploy directly to real infrastructure. This reduces dev‑prod impedance for agent-driven delivery and forces teams to rethink infra contracts, reproducibility, and runtime permissions — Principle 07.