Daidala AI provided 23 tasks to a 9-year-old child.

Daidala AI gives a company an autonomous team of specialized AI agents that carries work from a request through to a finished result, under the direction of the company's own people. How it works: every request becomes a tracked task. Daidala breaks the work down, routes each part to the agent whose role covers it, holds it to verification and review, and records every task, decision and change on the way. A person stays involved wherever human judgment matters, and can let a whole process run on its own when they choose. What that means day to day: an employee can lead a task that reaches beyond their current expertise, and an experienced specialist can run several workstreams at once while Daidala carries out the steps. The same way of working applies across different areas of a business. Software delivery is our best-documented example so far, because it is the area where the work has been measured: From July 10 to August 8, 2026, in our own company and across our own projects, the pipeline closed 928 tasks and merged 451 code changes. Median time from request to closed task was 39 minutes. About half of the tasks had no human step recorded in Redmine. This is one company, our own projects and one pipeline — not a benchmark against human teams. Daidala works on production tools: Redmine for coordinating the work, GitLab for code. Every task, decision and change stays visible and traceable there. Follow this page to see what autonomous AI work looks like in practice, what it delivers, and where its limits are. Book a walkthrough at daidala.tech.