Applications are now closedRegister for future cohorts →

London · One-week cohort

Gen Stream

Run atLISA·Supported byKairos

A one-week AI safety programme in London for aspiring generalists. Ship a real project and leave with evidence, connections, and next steps.

London · United Kingdom

Register interest

Applications for the August 2026 cohort are closed.
Register your interest for future cohorts.

Team

Gen Stream is being built by a small UK-based team focused on getting more capable generalists into AI safety work.

We design the week, run it in person, and stay involved with participants after it ends.

Yaqoob Ahmad

Co-founder & Facilitator

Yaqoob Ahmad

Runs the programme in person and facilitates the week, with a focus on design, shipping, and building the agency side of Gen Stream.

LinkedIn ↗
Julien Sireau

Co-founder & Facilitator

Julien Sireau

Runs the programme in person and facilitates the week, with a focus on curriculum, mentoring, and next steps for participants after it ends.

LinkedIn ↗
Al-Rayan Al Habsi

Operations Lead

Al-Rayan Al Habsi

Main point of contact for participant logistics and welfare, from accommodation and travel to venue coordination, guest speakers, daily feedback, and Demo Day.

LinkedIn ↗

Advisors

Our advisors bring deep experience across strategy, operations, founding, fieldbuilding, and management.

Sam Smith

Sam Smith

LinkedIn ↗Oliver Kurilov

Oliver Kurilov

LinkedIn ↗David Arnold

David Arnold

LinkedIn ↗

Applications

Closed

Register for future cohorts

When

Next cohort TBA

1 week, in person

Where

London

United Kingdom · in person

Programme size

8–12

Cohort of eight to twelve

Support

Funded

Travel, accommodation and food covered

The problem

AI safety's most-cited talent bottleneck isn't researchers. It's generalists: the operations, programme, talent, comms and fieldbuilding people who make organisations function.

Orgs receive hundreds of applicants per generalist role but can't cheaply tell who can do the work. Capable people cycle through courses and fellowships, accumulating context but no employer-legible evidence they can execute.

The work

Every participant picks a project before the week starts from a curated list of real problems sourced from AI safety orgs, then spends five days building it. The final day is demo day, in front of the cohort and the wider network. Before you leave, you'll have recorded a short demo video, written a one-page summary, sent both to at least one relevant person, booked a 1:1, and drafted a 30-day plan. This can serve as preparation for longer programmes like the Generator Residency or the MATS Founding and Fieldbuilding track, as well as generalist roles in AI safety.

How it works

One week. Real work. A clearer picture of where you fit.

PRE

Preparation

An opening call, a short reading list for shared context, and choosing your project from the curated list of real problems from AI safety orgs. Travel in the day before and settle into the hotel, so Monday starts with work, not logistics.

MON

Orient, then lock in

Arrive at LISA, set up, opening session and speed friending. The first work block is scoping: cut the project down to something you can actually finish. Pitch it to the room at 4pm, commit, then 1:1s and reflections before dinner.

TUE

Ugly first version

A short talk on networking your way into an AI safety org, then two long build blocks. The bar for the day is simple: by 5pm something exists that did not before, however rough. Two 1:1s each evening, including with the organisers.

WED

Red team day

A talk on building agency, then midpoint critiques in trios where someone else's job is to find the weakest assumption in your project. Midweek is the last honest moment to pivot, so pivoting is expected, not a failure. Afternoon goes back to building.

THU

Finish and package

Morning is the last real build block. The afternoon turns to packaging: record your video, draft your outreach, write the handover. Evening 1:1s are spent watching each other's videos and cutting the parts that do not land.

FRI

Send, then demo

Re-record if you need to, send the emails, finalise your 30-day plan. The afternoon is demos in front of LISA, then closing reflections and dinner. The send is the point of the week: the work leaves the room.

Where this leads

Learn reading list

Start with the reading list, why generalists matter and where the work is, so you arrive with shared context.

On-ramp Gen Stream

The one-week, in-person cohort. Build a real project and get it in front of people who can act on it.

Fellowship go deeper

Move on to longer programmes like the Generator Residency and MATS Founding & Fieldbuilding.

In-role the goal

A full-time generalist role, operations, fieldbuilding, comms, chief of staff, at an organisation working on AI safety.

Who should apply

You're more ready than you think.

The cohort is a mix of students, AIS group organisers, career transitioners, and more senior people looking to sharpen their practical skills. You don't need years of experience, and there's no upper limit. Strong projects, organising work, or self-directed initiative in one of these areas is enough.

01

Operations & systems

Workflows, automations, hiring pipelines, and the infrastructure orgs run on. Backgrounds in ops, admin, tooling, or systems-building.

02

Programme & project management

Turning vague initiatives into plans that ship. Backgrounds in project delivery, product, or coordination-heavy roles.

03

Talent & recruiting

Sourcing, evaluating and routing people well. Backgrounds in recruiting, HR, community organising, or admissions.

04

Communications & marketing

Clear writing, positioning, and branding for a field that needs it. Backgrounds in comms, content, PR, or design.

05

Events & community

Retreats, workshops and programmes people remember. Backgrounds in events, hospitality, or community management.

06

Founding & strategy

Scoping new projects and orgs the ecosystem needs. Backgrounds in entrepreneurship, consulting, or strategy.

What you'd work on

Find a real gap. Make something. In one week.

Examples of what participants take on. Each one comes from a real organisation or the open-problems bank, cut down small enough to finish in a week and still be worth using.

01

A talent CRM an org will actually maintain

Design and ship a lightweight candidate/stakeholder tracker for a fieldbuilding org, with the handover doc that keeps it alive.

02

An event runbook from a real retreat

Turn a past programme's scattered notes into a reusable runbook the next organiser can run without you.

03

A hiring rubric for generalist roles

Build a calibrated evaluation rubric for an org drowning in applications.

04

An onboarding SOP

Document and systematise how a small org brings people in, so the knowledge stops living in one person's head.

Applications closed · Future cohorts

Get involved with Gen Stream

Applications for the August 2026 cohort are closed. Register your interest for future cohorts as a participant, or contribute a real project as a contributor and shape what the next wave of AI safety generalists work on.

FAQ