Can enterprises trust AI agents? Pretxt bets $5m on it

Can enterprises trust AI agents? Pretxt bets $5m on it

Pretxt, an enterprise AI start-up focused on overseeing AI agents while they work, has come out of stealth with $5m in seed capital.

Its launch introduces what it calls AI Supervision, a new category of technology built to hold autonomous agents to an organisation’s own standards as tasks unfold.

The seed round drew backing from Deciens Capital, Howdy Partners, Service Provider Capital and NSV Wolf Capital. Pretxt plans to use the capital to establish runtime supervision as a dedicated layer of enterprise AI.

The aim is to let businesses hand agents more consequential responsibilities without exposing themselves to the failures that security and monitoring tools cannot stop.

The company argues that AI agents now do far more than produce content. They speak with customers, connect to other systems, process transactions and run business processes. Yet they do not always behave as instructed. An agent may skip a mandatory step, create its own exceptions, make commitments it has no authority to make or ignore an approval requirement.

A support agent offering an unauthorised refund, or a financial services agent bypassing identity checks, can trigger immediate operational, financial and reputational damage, with no malicious attacker involved.

Pretxt contends that security, governance, guardrails and observability each cover part of the risk but do not confirm that an agent is completing its assigned task to business or regulatory standards.

Some guardrails halt prohibited actions and leave processes stuck awaiting human intervention, while monitoring tools only show what already occurred. The stakes are clear: 83% of US consumers believe brands should answer for AI-driven interactions that go wrong.

Pretxt builds a runtime supervision layer that assesses agent behaviour within the context of each task, redirects agents that stray from approved processes and flags ways to strengthen oversight over time, with humans reviewing those changes. The platform deliberately avoids using LLMs as a judge, which the company says removes needless latency, cost and inconsistent decisions, delivering supervision 60 times faster.

Compliant interactions continue with millisecond-scale checks, while actions that breach an organisation’s boundaries are caught and resolved before reaching a customer or system of record.

In testing on Zapier’s AutomationBench, which mirrors real-world support workflows, Pretxt cut policy-violating agent actions from almost one in five to zero. The platform also learns from live agent behaviour, spotting gaps in guidance and suggesting refinements. Teams set expectations in plain language, test proposed plans against their own examples and approve updates before they go live.

Pretxt is aimed at high-stakes workflows, especially in regulated industries where an agent’s error can carry compliance consequences.

Pretxt co-founder and CEO Brian Soby said, “You don’t need a prompt injection attack for an AI agent to do something it shouldn’t. Security remains essential, but the next question is whether enterprises can trust agents to carry out real work according to their standards. You can’t supervise a workforce of AI agents with hope, a rulebook and reviews after the fact. AI needs supervision while it acts. That is the category Pretxt is here to establish.”

Pretxt co-founder and CTO Aaron Costello added, “Supervising an AI agent shouldn’t mean becoming an expert in everything it touches, or in how it’s built. Every platform has its own permission model. Every agent has its own tools and its own way of working. Nobody should have to master all of that before they can trust an agent with real work. Pretxt takes that complexity off the table.

“The agent already knows its job. Teams tell us, in plain language, how it should behave and where the boundaries are, and we supervise and steer the agent to stay within them. No untangling permissions, no digging into the agent’s internals. That’s a big win for trust, and for outcomes.”

Read the daily RegTech news

Copyright © 2026 RegTech Analyst

Enjoyed the story? 

Subscribe to our weekly RegTech newsletter and get the latest industry news & research

Copyright © 2026 RegTech Analyst

Investors

The following investor(s) were tagged in this article.