ALX™ · The Agentic Leadership Index
AI didn't break your leadership system. It just made the breakage impossible to ignore.
ALX measures one thing: how well you actually lead humans and AI together. Thirty questions, five pillars, 8–10 minutes, one number. Free for individual leaders — and the entire instrument is published in the open.
ALX v1.0 · Published July 2026
Most leadership assessments are black boxes: answer thirty questions, surrender an email, trust the score. ALX takes the opposite bet. Every question, the scoring model, the band cut-offs and the known limitations are published — versioned, citable, and free to scrutinise.
Why open? Because a standard you can't inspect isn't a standard. The questions were never the moat — the dataset is. Every completed assessment feeds the first published Agentic Leadership Index, coming January 2027.
Five pillars · weighted by leverage
Not personality. Not potential. Behaviour — what you actually do, in your own team, this quarter. The pillars are weighted: fluency is the foundation, so it carries the most; the weights are published, and they're a design judgement the data will be allowed to overturn.
Not whether you use AI tools — whether you lead with them, visibly, in ways your team can copy. Without this, none of the others fully function.
Whether work is designed across humans and agents on purpose — or by accident.
The speed of your decisions is the speed of your organisation. This measures your decision debt.
Whether you multiply what people achieve, or just supervise what they do.
Whether you learn faster than your environment changes. Everything else has a half-life.
The maturity ladder
Your score places you in one of four bands. Once the first index publishes, it will also place you against everyone else who has taken it.
Coming January 2027
One published report, every year: how leaders actually score, pillar by pillar, band by band — sample size and limitations stated out loud.
The field is full of confident claims about AI leadership and almost no published measurement. The index is the correction. Every individual assessment feeds it, anonymised. Organisations running ALX across their leadership bench feed it faster — and get their own benchmark back.
Writing
Not research — a field note. What the earliest ALX responses suggest, what they can't tell us yet, and why the honest answer to "what's your sample size" is the most useful thing in this piece.
What an exchange with Ethan Mollick clarified about the one input no agent supplies for you.
On Boris Cherny's archetypes, and why the shape of your team is now a design decision rather than an org-chart inheritance.
Advisory
The index is free. The work isn't. If your leadership population scores where early responses suggest most do, the gap is orchestration — and that's a design problem, not a training problem.
Agentic leadership strategy, manager capability, and AI-enabled operating models — designed inside the organisation, not sold in from outside.
Run the index across your leadership bench and benchmark against the full dataset as it grows. Teams from £500. Enterprise from £20,000 a year. Founding pilot organisations run free in exchange for anonymised data.
The instrument, the fieldwork, and what it means for the people in your room.
About
Every framework here was built and tested inside a real organisation. None of it was designed in a consultancy and sold in.
From 2020 to 2025 I led learning transformation at the Chan Zuckerberg Initiative — through a pandemic, and alongside scientists trying to end disease. I now lead agentic talent experiences at a global technology company, which means I spend my days doing the thing this index measures. Before that: Google, Kayak, OpenTable, Agoda, Groupon, News International. Three continents, twelve organisations, twenty years.
The index exists because I could not find one I trusted.
— Glyn Burleigh-Sheard