Gen Stream · London, UK
A one-week, in-person intensive that turns ambitious junior generalists into contributors in AI safety.
London · United Kingdom
The next cohort is forming. Register for the next round. [COHORT DATES TBC]
AI safety’s fastest-growing talent need is no longer more researchers. It’s competent generalists — the people who build organisations, run programmes, and hold projects together.
Plenty of ambitious junior people want to help make AI go well. What’s missing is a legible, low-commitment way in — somewhere to build real experience and credibility before competing for scarce full-time roles. That gap is widest at the top of the funnel, and for talent in the UK and EU.
Gen Stream is that on-ramp. In one focused week you go from understanding the problem to having done real generalist work on it — with mentorship, a tangible output, and concrete next steps into the field.
The next cohort is forming. Register your interest and we’ll let you know when applications open.
You should apply if you want to contribute to AI safety but suspect your comparative advantage may be execution rather than research.
We’re especially excited about two groups. First, student or recent-graduate AI safety university group organisers who have strong context and mission alignment but want more professional experience. Second, career transitioners with strong execution skills from operations, project management, events, consulting, startups, recruiting, communications, executive assistance, education, or community building — even if you’re still building AI safety context.
You don’t need to code. You do need to be reliable, curious, clear in writing, comfortable with ambiguity, and unafraid to sometimes do unglamorous work that helps important projects succeed.
Good fits may have done things like:
Probably not the right fit yet:
We look for general competence, good judgement, and genuine thoughtfulness about the risks from advanced AI — more than we look for credentials or a particular background. If you’re unsure whether you’re a fit, err towards applying — we’d rather read your application than have you rule yourself out.
Gen Stream is early-stage, and a cohort may not run if there isn’t the right level of interest or fit. Accepted applicants may be offered a place in a later cohort. If you need a decision by a particular date — a visa, a competing offer — tell us and we’ll do our best to work to it.
Built by early-career generalists who hit this exact gap.
Gen Stream complements programmes further along the funnel rather than competing with them. Generator Residency, for example, is an excellent mid-to-late-stage programme — but it’s US-based, runs once a year, and is built for people who are already experienced. That leaves the top of the funnel, and most UK and EU talent, underserved.
Our aim isn’t to take already-excellent generalists and polish them. It’s to repeatedly take capable, ambitious people who are early in their journey and give them the experience, connections, and next steps to genuinely enter the field — and to do it often enough to matter.
Generalist work spans a lot of ground. A few directions a cohort might take — illustrative, not a menu:
Design and run something that brings new people into AI safety — a workshop, a reading group, a local event — and learn what actually moves people from interested to involved.
Embed with a small organisation and build a system it needs to scale: a hiring pipeline, an onboarding flow, a process that was quietly eating someone’s week.
Scope and prototype an event for a specific, high-leverage audience — who to bring together, what it’s for, and how you’d know afterwards whether it worked.
Take a hard idea in AI safety and make it land for a particular audience: an explainer, a briefing, a piece of writing aimed at people who don’t yet care.
Help an organisation reach people it currently can’t — sourcing, work tests, referral routes — and document what you learned so the next team can reuse it.
Map a corner of the field that’s poorly understood — the gaps, the open questions, who’s working on what — and turn it into something other people can use.
Pressure-test an idea for something that doesn’t exist yet: who it’s for, why now, and the smallest version you could ship to find out whether it’s real.
These are starting points, not a list to choose from. Most of the value comes from finding a real gap yourself and making a credible first move on it.
Not a lecture series. You spend the week doing real generalist work on a real problem, with the scaffolding to make fast, visible progress.
Time with people working inside AI safety organisations — on your project, your questions, and how the field actually fits together.
You leave with something concrete you made: evidence of what you can do that you can show to anyone deciding whether to back you.
Introductions, a clearer map of where you fit, and a route into the programmes and roles that come next.
Gen Stream is intense and output-driven. You’ll spend part of the week learning how high-bar AI safety generalists work, and most of the week doing the work yourself — meeting professionals, scoping real problems, completing individual work trials, building something useful for a partner organisation, and leaving with a clearer sense of your fit.
Learn the generalist landscape, meet the cohort, map your skills, and hear real problems from AI safety organisations. Leave with: Fit Map v1 and shortlisted problem briefs.
Complete an individual work trial using tools like Sheets, Notion, Airtable, Claude, and docs. Leave with: an evaluated work sample.
Try different generalist tasks, interview problem owners, and submit a short video application exercise. Leave with: a profile evidence matrix and capstone brief.
Work in a small team on a real capacity constraint from a partner org or hub. Leave with: a capstone artifact and portfolio draft.
Present your work, receive evaluator feedback, finalise your portfolio, and plan what’s next. Leave with: final portfolio, profile recommendation, and a 30-day plan.
Start with our short, self-paced course (~90 min): why generalists are the binding constraint, why the pipeline is broken, and what you can do about it.
The one-week, in-person cohort. The natural next step after the course — and the stage most people skip. This is it.
Move on to competitive, longer programmes like Generator Residency that take you further into the field full-time.
A full-time generalist role — operations, fieldbuilding, communications, chief-of-staff — at an organisation working on AI safety.
Still have questions? Email us at [EMAIL TBC].
Tell us a bit about you and we’ll be in touch when applications for the next Gen Stream cohort open. Places are limited.