Pretxt AI Launches Supervision Tool With $5M Seed

The startup's AI Supervision platform monitors autonomous agents at runtime against company rules, backed by a seed round led by Deciens Capital.

Updated on Nov. 17, 2026 in Artificial Intelligence

Bold vector editorial illustration of a brass valve regulating a steel pipe, a concrete image of automated constraint on a running system.
Pretxt AI launched its AI Supervision tool and $5 million in seed funding led by Deciens Capital, letting companies halt autonomous agents that break their rules. AI Illustration. Upload story photo >

Pretxt AI launched its AI Supervision tool alongside $5 million in seed funding led by Deciens Capital, with participation from Howdy Partners, Service Provider Capital and NSV Wolf Capital. The platform monitors AI agents at runtime against a company's rules and policies and intervenes when an agent begins to drift off course.

Why it matters

Autonomous agents have shown erratic behavior, including hallucinated responses and skipped workflow steps, while existing security tools either block agents outright or flag errors only after the fact. Real-time guardrails are likely to become a priority for enterprises deploying these systems across the United States.

The platform evaluates agent actions in milliseconds using a near-deterministic engine rather than a second large language model as judge. The company's reported figures come from its own tests on Zapier's AutomationBench, and it has not said how the speed claim was independently measured.

The players

Pretxt AI

Pretxt AI is an artificial intelligence startup that develops runtime oversight tools for autonomous AI agents.

Deciens Capital

Deciens Capital is an investment firm that led the seed funding round for Pretxt AI.

The details

Users select which policies, standards and guardrails apply to an agent and track them in real time. When an agent is about to break a rule, the platform halts it and escalates edge cases to a human decision-maker. The AI Supervision tool is available in early access.

Timeline

  1. • November 17, 2026: Pretxt launches the AI Supervision tool and announces its seed funding.

The Tech Race

Pretxt's approach departs from the LLM-as-a-Judge designs that many AI security tools rely on, offering a deterministic alternative the company says runs up to 60 times faster. The launch places a new entrant in the contest to govern autonomous agents before they are widely deployed.

For businesses running AI agents on customer or internal workflows, runtime guardrails could reduce policy violations without requiring every action to be reviewed by hand. The tool remains in early access, so most organizations will not be able to adopt it immediately.

The takeaway

Organizations deploying AI agents should consider how they will monitor agent actions against their own rules as the systems take on more autonomous work. Testing any supervision tool against your specific workflows before relying on it is the practical next step.

Further reading

Explore more coverage in Artificial Intelligence for context on how autonomous agents are being governed.