Context

Inside Deloitte Canada’s Supply Chain & Industry Operations team, our principal led a business-development analysis with one instruction worth respecting: no magic, only use cases someone could actually build and buy.

The problem

Supply chain is drowning in AI pitches and starved of AI deployments. The gap is almost never the model — it is feasibility (does the data exist?) and economics (does the payback close?). The mandate was to find the intersection.

What we built

The filter started with people: structured interviews across the division’s stakeholders to locate genuine operational pain. Candidates were screened for data availability and integration reality, not leaderboard scores.

Five use cases survived — each documented with a high-level solution architecture and adoption roadmap, and each run through capital-recovery analysis under a SaaS delivery model with a 40% ROI benchmark. Feasibility and finance travelled together, deliberately.

Key decisions

Killing good ideas early. The portfolio’s value came from what was excluded: technically sweet cases with no data path, and financially tempting ones with no operational owner.

Outcome

A five-use-case AI portfolio with architectures and payback models attached — and the advisory discipline JES still uses when a client asks, “where should AI go first?”