AI helped design a modular “kit” robot that can keep walking after losing a leg—cool demo, but the real test is rough terrain, long runs, and repeatability.
Solo.io unveiled Agentevals at KubeCon EU 2026 to score AI agent workflows using telemetry—so teams can compare versions, catch regressions, and control cost.
Oracle’s adding autonomous AI “agents” to Fusion Cloud to run real workflows—not just chat. Big upside, bigger risks: permissions, audits, security, and trust.