About AgenticCareers.co

The agentic economy represents a new era where AI agents work alongside humans to accomplish complex tasks autonomously. We exist to connect the people building this future with the companies that need them, while tracking machine-executable work only when sources expose current, verifiable tasks.

This board exists because the general job boards handle this market badly. Search them for agent roles and most of what comes back uses AI terminology loosely: a backend position that mentions machine learning once, a data role retitled during a hiring push. The listings that genuinely involve building autonomous systems are scattered across a dozen different titles, because nobody has agreed on what to call this work yet. The same job is advertised as AI engineer, agent engineer, LLM engineer and forward deployed engineer depending on who wrote the posting.

So we do the sorting. Every listing is pulled from the source, scored for whether the work is actually agentic, categorised by what the role does rather than what it is called, and pruned when it closes. Roles that turn out to be neither agentic nor current do not stay on the board.

An automated pipeline runs every morning across company career pages and applicant tracking systems including Greenhouse, Ashby, Lever, Workday and SmartRecruiters, alongside a set of specialist sources. Each listing is deduplicated against what we already hold, scored for relevance, and dropped if it is not in English or no longer open at the source.

A publication gate sits at the end of that run. If any single source returns dramatically fewer roles than the day before, the whole update is rejected and the previous dataset stays live rather than publishing a corpus with a hole in it. That gate trips every few weeks, usually because one source rate-limited our crawler, and it is the reason the counts on this site move gradually rather than lurching.

Everything we publish is measurable from that corpus, and where a slice is too small to support a percentage we say so instead of quoting it. The underlying data is available as an open CC BY 4.0 dataset on the data page, and as JSON at /api/jobs for anyone who wants to check our work or build on it.