Monte Carlo launches Bayes to find and repair faults in AI agents

Monte Carlo, a company that sells tools to monitor data and AI agents, launched Bayes on 28 October 2026. An AI agent is software that works without a person. Bayes finds faults in agents and in their data, and then repairs them. Monte Carlo says Bayes found 200% more incidents than the systems that the customers use, with incorrect alerts down by more than 80%.

How it works

Bayes learns the data and AI systems of a customer. It then sets monitor tools. It checks the output of agents, their decisions, the context they get and their data. The customer selects how much Bayes does without a person.

The numbers

Monte Carlo says that one engineer can monitor only one to three agents. A Deloitte study found that 70% of leaders are not sure that they can control their agents. Monte Carlo gave no test method for the 200% and 80% numbers.

What is not known

The company gave no cost and no test by an external group. Customers of Monte Carlo can use Bayes in the platform, in Slack and Microsoft Teams, and through MCP. MCP is a connection that lets AI tools use other software.

Insight: The agent that monitors agents also needs a record

Monte Carlo says that engineers can monitor only one to three agents, and it sells an agent to monitor agents. A Nylas study on 20 October 2026 found that only 53% of firms can fully check the steps of their agents in the last week. This reads as a move to control the layer that monitors agents. That layer is one more agent that must have a record. Get that record too.