Workflows: conveyor and ship-loader stoppage response, maintenance work-order planning, control-room shift handover and incident reporting, remote pipeline leak inspection, and permit-to-work and isolation checks.
This is an illustrative example. Saltbush Bulk Terminals is a fictional company; the ratings, reasons, and value ranges below are Use Case Foundry's output for its company model, not claims about a real business. Values are the engine's estimates in US dollars.
The top three, scored
| Opportunity | Value | Practicality | Company advantage | Evidence quality | Estimated annual value |
|---|---|---|---|---|---|
| Compliance evidence copilot | High | Medium | High | High | $710k–$1.8M (upper bound) |
| Maintenance planner copilot | High | Medium | Medium | High | $130k–$325k |
| Shift handover and incident copilot | High | Medium | Medium | High | $60k–$150k |
Value ranges are estimated from the recorded cost of the pain each opportunity addresses. The compliance copilot applies across the terminal, so its range is sized against all five recorded pains combined, about $3.55 million a year; treat it as an upper bound until a pilot measures it.
1. Compliance evidence copilot
- Why it ranks first: critical-infrastructure security obligations apply to plant systems and data, and safety isolation and permit decisions must stay with authorised people. The regulatory constraint applies across the whole terminal, not to one workflow.
- Company advantage: compliance capability is itself a barrier, and the copilot is grounded in the company's own records.
- Before a pilot: the copilot assembles evidence and drafts findings; an authorised reviewer decides every case, and it does not change plant settings or approve an isolation. Use Case Foundry has a tested pilot path for evidence assembly.
2. Maintenance planner copilot
- Why it ranks second: planners build work orders by hand from scattered history, so parts and crews arrive late for shutdown windows, a recorded pain of about $650,000 a year.
- Company advantage: it captures scarce planner judgment and uses the company's own work-order history.
- Before a pilot: the planner approves every work order; test the copilot against past shutdown windows before it is used on a live one. Use Case Foundry can produce the pilot package; there is no tested production path for this pattern yet.
3. Shift handover and incident copilot
- Why it ranks third: handovers are verbal and incident reports are inconsistent, so recurring faults are rediscovered each shift, about $300,000 a year.
- Company advantage: it is grounded in the terminal's own shift logs and incident reports and in control-room operator judgment.
- Before a pilot: the copilot drafts the handover record and the outgoing operator confirms it; measure whether recurring faults are caught sooner.
Why the largest pain is not the first AI pilot
Unplanned conveyor and ship-loader stoppages are the largest recorded pain, about $1.8 million a year. Use Case Foundry still marks that workflow "instrument first": sensor coverage is partial, and the condition data that exists stays on the plant network, not accessible for analysis. Physical automation for stoppage response ranks 13th, with low practicality and low company advantage, and needs a specialised engine or model that the pilot package does not ship.
Remote pipeline leak inspection follows the same pattern: a $550,000-a-year pain, with inspection records on paper. In both cases the first investment is capturing data at the site, not building a model.
What this means for infrastructure and edge teams
Use Case Foundry does not size compute or select hardware. It gives an infrastructure team the business reason and the order: which AI work is worth running, which site data must be captured first, and the reliability and safety limits the solution must respect. Here, the two instrument-first items carry about $2.35 million a year of recorded pain, which is the business case for site-level data capture before anyone sizes a model.
Why automation waits
Helping the person ranks above automating the work in every workflow: maintenance planning 2nd against 11th, shift handover 3rd against 12th, and permit and isolation verification 8th against 23rd. Every automation option depends on evaluation and monitoring guardrails, ranked 10th, because a wrong automated action on live plant is a safety risk.
Why integration is partner work
Connecting AI to legacy SCADA and the plant historian is a real opportunity, but Use Case Foundry does not treat interface mapping as generic software. The roadmap marks it as implementation-partner work: the delivery team maps the plant systems, network boundaries, and security controls after the opportunity is selected.
Buyer question this answers
Which AI should an industrial or critical-infrastructure operator fund first across remote sites, what must be instrumented before AI can address the largest operational pain, and where must authorised people stay in control?
This ranking starts from internal operating evidence. See how a first read is built from public information alone in the sample pre-assessment.