theona.ai
theona.ai turns the way a company already works into agents that run it: the platform maps the real process, then builds agents on top of that map. It is aimed at companies from about a thousand people, where the processes already exist. I worked between the product and the people whose work it changes. Full-time from January to September 2026, then on contract until November.
- My job was to fit agent systems into the everyday work of a team, so that the processes they already had got stronger.
- Built the agent systems behind the team's public communication about the product: research, drafts, fact checking, critique, filters that catch AI slop, visual content generated on top of the team's design system, and the analytics on what gets published.
- Worked directly with the CEO on founder-led content: the strategy, his LinkedIn and Substack, and a drafting agent that writes in his voice, learned from his own edits rather than from a style guide.
- Eight people used those systems every day. I wrote the rules for working with them and coached the team until they ran them without me.
- Put measurement where there had been none: a weekly tracker, per person and per post. Before it the company did not write publicly at all and had no way to see whether any of it worked.
- Taught HR, sales and operations specialists to build agents for their own work. I prepared almost every theona workshop and demo and ran several of them: a corporate training where a client's HR team built agents on top of their own hiring system, the materials and demo agents for an executive workshop run by the group's CEO, a facilitation workshop for HR practitioners, and the demos for HR Tech Europe in Amsterdam and Web Summit Vancouver. For each one I built the demo agents myself, set up test environments with realistic data and a reset script so every group started clean, and wrote a participant guide with ready-to-use prompts. People left having built an agent, not having watched one.
- Owned the use-case hubs on the product site. I mapped the hiring, sales and operations funnels step by step, decided which step an agent takes and which one stays with a person, and turned each step into a use case the platform builds an agent from in one click. Most of the 65 published use cases are mine, and I tested each one to make sure its agent actually runs.
- Turned that audience into pipeline: a chain that reads every reaction and comment on the team's posts, filters out our own people, scores the rest against the ICP and hands them to the CRM, with a weekly digest in Slack. 29% of the scored contacts came out high priority against the ICP.
- Wrote the case study of theona's pilot with the Agentic Builders fund, interviewing the managers who used it about what changed in their work.








