Where the next power list is coming from — the patterns and pipelines producing very young agent builders.
| # | Who / what | Why watch |
|---|---|---|
| 1 | Student builders shipping MCP servers | University-age contributors dominate the protocol's tool ecosystem |
| 2 | Hackathon agent teams | Weekend agent projects increasingly become funded startups |
| 3 | Open-source maintainers under 25 | Core contributors across agent frameworks and eval suites |
| 4 | AI society founders on campus | University AI clubs now run agent demo nights, not just reading groups |
| 5 | Young eval researchers | Benchmark authorship skews younger every year |
| 6 | Indie agent-tool makers | Solo builders selling agent infrastructure as micro-SaaS |
| 7 | Robotics students | Embodied-agent work pulling fresh talent into the field |
| 8 | Content educators on YouTube/TikTok | The under-25 teachers of the agent stack to their own generation |
| 9 | Competition winners (Kaggle, olympiads) | Direct pipeline into frontier labs' agent teams |
| 10 | Teen open-source contributors | Framework PRs from contributors still in school are now routine |
Part of the Engatica rankings — see the AI Agents Power List 2027.
A watch list of the next generation of agent builders — and of the patterns and pipelines producing them: hackathons, open-source projects, university labs and founder programmes where very young builders are shipping real systems.
No. The list focuses on builders early in their careers and on the routes that produced them, rather than applying an arbitrary birthday test.
Yes — nominations are open. Point us to shipped work: a repository, a product, a competition result. Potential is interesting; evidence is persuasive.