AI systems and judgement
AI ecosystems, reinforcement learning, engineering and development, model capabilities, governance, safety and quality.
Executive learning / applied AI / commercial operating systems
AI should surface judgement, not absorb agency.
I bring the unique combination of technical fluency and the operator's judgement to know when AI should improve an outcome, how to measure it and how to make the intervention accountable.
Explore the Applied AI Program ↗
01 / The programme
From AI curiosity to operating capability.
MasterClass Executive is a selective, compressed AI leadership programme with an MBA-style curriculum, created with the University of Chicago Booth School of Business and built in collaboration with OpenAI. Its intensive approach combines business education, adaptive learning, expert-led sessions and hands-on labs.
I am an ambassador member of the founding cohort. The founding cohort comprised 700 members selected from 29,000 applicants.
This is an executive-learning programme, not a Chicago Booth MBA degree or an OpenAI employment credential.
700founding-cohort members
29kglobal applications
Ambassadorfounding-cohort member
02 / What I am learning and building
An applied curriculum spanning strategy, systems and execution.
Investment and use cases
Financial analysis and modelling, customer research, synthetic use cases, use-case selection and return-on-investment design.
Prototypes and agents
AI prototypes, agentic workflows and proprietary agents for sales and go-to-market execution.
Commercial applications
Answer-engine optimisation, customer-support automation, storytelling and slide decks with AI, and learning to work with AI itself.
03 / AI in GTM · the judgement opportunity
THE QUESTION
So, what have we automated?
AI adoption is not the same as commercial return. The test is whether it removes work sellers cannot get to, exposes the evidence that matters and helps a person make a better decision.
ADOPTION HAS ACCELERATEDSELLING TIME HAS BARELY MOVED
20182026
Sales organisations using AI
21%87%
+66ppSeller time spent selling
34%40%
+6ppThe gap is the point: more AI use has not, by itself, produced a comparable shift in time spent selling. These are directional comparisons across separate survey editions, not a causal ROI calculation. Sources: Salesforce State of Sales, 4th edition and 2026 findings.
THE DEFAULT
Automate activity
ResearchMessagingOutreachFollow-up
More output enters the system. The seller still has to reconstruct the account, separate signal from noise and decide what to do.
THE BETTER DIVISION OF LABOUR
Prepare judgement
Join dataClear adminRank changeExplain evidence
AI handles the evidence burden. A human retains the relationship, the intervention and accountability for the decision.
01 / AIJoinFragmented evidence
02 / AIDetectMaterial change
03 / AIRankAttention, not volume
04 / AIExplainReason and source
05 / HUMANDecideAccept · adapt · reject
ACCOUNTABLE DECISION WORKSPACEIllustrative / synthetic data
01 / WHERE TO LOOKRanked attention
01Renewal risk surfacedBuying-group engagement has changed
02Expansion signalProduct use is rising in a new team
02 / WHY IT MATTERSInspectable evidence
What changed? Which customer behaviour supports it? What evidence is missing?
03 / WHAT TO DOSeller-owned action
Review the evidence, choose the intervention and record the reason.
WHAT BREAKS RETURN
Faster activity can mean faster leakage.
Broken process+Amplified activity→Optimised burn, not growth
Too many alerts recreate noise. A CRM stage is not buyer evidence. A fixed play ignores context. The system must be able to say “not enough evidence.”
A BETTER FIRST TEST
One opportunity. Three material signals. One accountable seller.
Start with a live or anonymised opportunity. Make each signal inspectable and falsifiable. Let the seller accept, correct or reject it. Measure whether it surfaced something worth acting on before expanding the workflow.
Evidence boundary: the decision workspace above adapts a synthetic prototype. It illustrates a method; it does not claim measured conversion, productivity or revenue gains.
04 / Learning environment
Learning from people shaping the field.
Programme and event learning draws on leaders and organisations working at the frontier of AI products and business adoption.
Builders and companies
Clay · Decagon · ElevenLabs · Gamma · GraphAI · OpenAI · Weights & Biases · Micro1
Instructors and speakers
Grant Lee · Mark Cuban · Indra Nooyi · Ray Dalio · Gary Vaynerchuk · Jerry Dischler · Steven Kaplan
05 / The operating test
Where can AI change the economics?
My starting point is never “Where can we add AI?” It is: which decision, workflow or customer experience is constrained, and what evidence would show an AI intervention has improved it?
- Define the economic problem. Identify friction in conversion, retention, capacity, quality or speed.
- Choose the right intervention. Compare AI with process, product, data and human alternatives.
- Build and adopt. Prototype, test, put accountable owners around the workflow and enable the team.
- Measure the result. Track return, customer outcomes, adoption, safety and governance.
That is the combination I bring to a commercial mandate: technical literacy, commercial operating experience and the discipline to separate a promising demonstration from durable value.