AI Opportunity Discovery

Why discovery is the only leverage point, and the constructs everything else rests on.

1. What happens before opportunity discovery. Why your discovery process has to differ from the one the business already uses · meet the business where it is · what is in it for them · optimize for constraints, maximize ROI · the first-principles definition of AI value creation, taught with zero technical content · the cheapest-viable-technology rule · bolt-on against re-engineered monetization
2. A new mindset and understanding. Anatomy of an Insight, a workshop that reverse-engineers a finished insight to expose everything hidden beneath it · opportunity pipeline thinking · the maturity model, introductory pass · workflow change as intervention, the unit of analysis for the rest of the course

Cases A hyperscaler that failed to see the true size of the cloud opportunity and went cloud-second as a result · second-generation cloud and ads products as bolt-on failures at scale · retail list discipline and generative search as the intervention

Exercise Map one workflow in your business end to end. Identify where technology is currently used intentionally, and where it is not.

Where the money is, before you go looking for opportunities. Skipping this unit tends to produce technically sound opportunities with no monetization path.

1. Data as an asset. Four competitive-advantage criteria for screening opportunities · reframing the business as a data-generating entity · two monetization modes, direct and aligned · the alignment guardrail
2. A new paradigm of monetization. Ecosystem business models, where AI platforms have partners rather than only customers · the scaling-access shift · creativity as the new productivity, and the AI economy as an optimization economy · knowing something about the market no one else knows
3. Partnership monetization. The partnership opportunity checklist · the myth of might be · adjacency analysis · adversarial opportunity discovery, and when the correct output is a response plan rather than an initiative

Cases A forum business cutting off free scraping and licensing its archive · a studio treating intellectual property as data · a ride-hailing platform turning drop-off location into ad targeting at the point of opportunity · two model-lab partnerships, one of which failed · an enterprise software vendor moving from walled garden to consumption credits to a partner ecosystem · consumer negotiation agents as a coming disruption

Exercises Re-engineer a media-consumption workflow so it generates data no other medium can produce · extend the scaling-access pattern to your own domain · re-engineer insurance discovery and claims on the assumption that you are partnering with the customer

Top-down, bottom-up, and the on-the-fly progression.

1. The opportunity discovery frameworks. The three-pillar construct of business model, operating model, and technology model · top-down discovery and the four hype-resistant questions: is the technology ready, is the business model ready, is it feasible for us, are we too late · bottom-up discovery as AI product governance, vetting frontline ideas for ROI before they reach technical teams · shelve rather than say no · assumption budgeting per release
2. Real-world opportunity discovery. The four-point progression for running discovery live, or in a hallway: what is the critical KPI, why this technology, what is the recommendation, what information advantage is this built on. Plus downstream breakage analysis, and trust as a precondition for big recommendations

Cases Two streaming platforms and their creator AI tools, taught deliberately as a failure case, because correctly killing an attractive opportunity creates value faster than pursuing it · retail margin compression reframed from layoffs to growth

Exercise Replace the worked case with your business and your biggest rival. Run all four questions. Then: how would you counter let us use AI for productivity and reduce headcount with a growth-centric response?

Feasibility and estimation, run immediately after a discovery session and before roadmaps, estimates, or commitments.

1. Opportunity feasibility assessments. Problem space: translate the opportunity into something buildable without prescribing how it gets built, and define success in business KPIs rather than model accuracy. Data space: is there low-cost access to a data-generating source, and synthetic data's hard limit, which is that it amplifies signal already present and cannot create new signal. Solution space: trust the team, ask rather than tell, listen for hedge language, ask for multiple candidate solutions
2. The opportunity estimation framework. Why single-number estimates get rejected and ranges are the native language of the C-suite · the three bands, where underperform is roughly 95% certain and has to carry the initiative by itself, expected is roughly 80%, and outperform is close to even and included so you are ready for it · old workflow to new workflow as the estimation mechanic · why being wrong is acceptable and being directionally wrong is not

Cases A multibillion-dollar model-lab investment reconstructed from latent pricing power in an office suite, where the actual runaway product was not the predicted one · a breakout mobile game as the argument for planning the outperform case · a theme park app as the most rational mixed-reality product available today · AI in a coffee maker as a failure case

Exercise Produce a three-band estimate for one opportunity in your pipeline. Test whether your underperform band alone would carry it.

Pragmatic futurism, paradigm spotting, and capital-I innovation. Lesson 10 is named the most valuable sub-framework in the course.

1. Pragmatic futurism. The four phases: disruption, where a new technology breaks a specific nameable assumption, then opportunity, then customer, then product. If you cannot define something buildable, pass. Four taught disruptions: the AI search paradigm, the death of reporting, time travel, and local AI. Plus why incumbents do not innovate until a startup forces them
2. Finding opportunity paradigms. Reading the tea leaves: listen for transition language, apply the money test, treat stated constraints as paradigms too. Worked example: data scarcity and compute scarcity, and information asymmetry as a monetizable product
3. Innovation opportunities. Capital-I innovation changes an assumption sitting underneath many products. Innovation creates new behaviors, so the adoption journey is part of the opportunity. If you cannot define an adoption journey, the opportunity is not real

Cases A decade of assumption to opportunity to customer to product at a semiconductor company, from exiting the CPU race through parallel compute, inference, edge, autonomous vehicles, robotics, and simulation · the comparative set, where the Walkman, the iPhone, and the action camera succeeded while camera drones and the metaverse failed on the same surface ingredients · a January 2026 industry keynote worked live

Exercise Work the AI-search disruption through all four phases to a buildable product for websites whose search traffic has collapsed.

The back-pocket toolkit. You can set everything up perfectly and it will still go badly. The recurring move: find the root cause, then redirect, without ever directly refusing.

1. The silent room. Restate the mission, review business goals, prompt with their pain points, keep ideas in your back pocket, and do not take over
2. Prescribing technical solutions. A trust problem rather than a discovery problem, and it takes three or four quarters of delivery to fix
3. We just need to do something with AI. The Four Rs: reassurance, root cause, redirect, restatement
4. The I-want-it-all problem. Accelerate and redirect, then pivot to what we can do and what we should do
5. The impossible external roadblock. Sometimes regulation genuinely justifies shelving. It is a failure mode only when one roadblock blocks everything
6. There is not enough data and none of this works. Their evidence is real and the conclusion is wrong
7. Prioritization by squeaky wheel or by job title. Challenge the premise and ask for proof, or let other organizations do it for you by surfacing displaced revenue

Capstone Carry one real opportunity from your own business through the full system: connect it to a critical KPI, justify AI over cheaper alternatives, source and frame it, identify its monetization paradigm and information advantage, run the three-space feasibility assessment, produce a three-band estimate whose underperform case carries the initiative, name what breaks downstream, and defend its priority in a pipeline.