Mate publishes CPR, its gate for autonomous actions

Mate Security published a post on 2026-10-18 describing CPR, short for Confidence, Precision and Reversibility. Mate says an agent's action is scored on the three before it acts alone, and that the platform enforces the scores outside the agent's reasoning loop. The platform page lists three autonomy modes: Recommend only, Supervised and Auto-run.

The post also says generic AI was right about 60 percent of the time on real customer cases and about 97 percent with Mate's context. Mate gives no sample size, so the figure is a vendor claim, graded C. A webinar on the framework is set for 2026-11-17 with a Latio founder, who also co-hosts an earlier Mate webinar.

Source: mate.security, read 2026-11-03.

What it changes for a buyer

Mate's approvals row stays at 8.5 for documented controls. The 97 percent figure is not scored, and CPR is described, not independently tested.

Other dispatches

• Mate Labs publishes the Blue City benchmark, without Mate in it, 3 November 2026
• XSOAR 8.11 reaches end-of-life on 2026-11-13, 3 November 2026
• Sentinel leaves the Azure portal on 2027-03-31, with automation differences listed, 3 November 2026
• ThreatConnect's domain now redirects to Dataminr, 3 November 2026
• Google updates its SecOps playbooks screen documentation, 28 October 2026
• Torq says IDC named it a Leader in its AI SOC MarketScape, 20 October 2026
• Google SecOps curated rules move to the GTI_IOC feed, 12 October 2026
• Splunk SOAR 8.7.0 drops Python 3.9 and adds a connector agent, 2 September 2026
• Mate Gamebooks reach general availability, 26 August 2026
• Microsoft updates its Splunk SOAR to Sentinel migration guide, 12 August 2026
• ServiceNow press report puts a Tier 2 SOC AI Specialist in December, 5 August 2026

More notes in Dispatches. To see the platforms side by side, start from the card.