Eon's Era generates test companies from real enterprise data structure
Eon released Era, a free public platform that builds virtual companies with Salesforce, Zendesk, Slack and Jira based on industry, size and tech stack, letting agents run tests on realistically messy data that can be reset and rerun.
In 2025, Salesforce researchers put leading AI agents through realistic CRM work. They succeeded about 58% of the time on single-turn tasks and around 35% once the task became a multi-turn conversation. And a benchmark is still tidier than a real company. A working company is messy in specific ways. The same customer sits in the CRM twice, and the Slack escalation contradicts the ticket. The APIs throw rate limits and vendor error codes that most test setups skip. Asking a model to invent a test company has the same flaw. It hands back data that is tidy and self-consistent, and real companies are neither. Their mess shows up at particular rates and in a particular order, and learning those rates takes a very large sample of companies. That idea sits behind Era by Eon, which is live, public and free as of today. The company behind it protects close to an exabyte of enterprise data, and Era generates whole companies using only the structure of that data. Pick an industry, a size and a tech stack, the company says, and Salesforce, Zendesk, Slack, Jira and more come back telling one story over vendor-compatible APIs. The open question is transfer. I'd still ask whether a score earned in a generated company predicts how the agent behaves in a real one. I'd expect teams shipping agents to lean hardest on the reset, which puts the same company back to its starting state. Change the model or the prompt, rerun the same tasks, and compare the two runs. Every company's data disagrees with itself somewhere. An agent should run into that in testing before it runs into it in production.
This post quotes another post: “Era goes live: a complete simulated enterprise for agent testing”