AI Pilot Scoping Canvas for Manufacturers

{ "Title": "AI Pilot Scoping Canvas for Manufacturers", "IntroductionHtml": "

Use this one‑page canvas to quickly capture whether an AI pilot addresses a real operational problem, can be evaluated safely, and is likely to deliver measurable value. Complete the sections below with practical detail a cross‑functional team can act on. Save a copy for governance, reuse, and follow‑up experiments.

", "SubmitLabel": "Save Canvas", "SuccessMessage": "Canvas saved. You can return to edit or export this pilot scope.", "DataType": "AIPilotScopingCanvas", "SchemaVersion": 1, "Fields": [ { "Key": "problem_statement", "Label": "Problem statement & user story", "FieldType": "textarea", "HelpText": "Describe the operational problem in plain language and capture a short user story: who is affected, what they need, and why it matters. Example: 'Line operator needs early warning of tool wear so we can avoid unplanned stops and scrap.'", "Required": true }, { "Key": "business_benefit_summary", "Label": "Expected KPIs & business impact", "FieldType": "textarea", "HelpText": "List the KPIs the pilot will influence (OEE, downtime minutes, scrap %, cycle time, throughput, on‑time delivery) and a rough expected impact or estimate (e.g., reduce downtime by 20%, save $40k/year). Be explicit about the metric and baseline where possible.", "Required": true }, { "Key": "estimated_financial_impact", "Label": "Estimated annual financial impact (optional) — USD", "FieldType": "number", "HelpText": "Enter a best‑effort dollar estimate of annualized savings or revenue uplift if the pilot succeeds. Leave blank if unknown.", "Required": false }, { "Key": "data_sources", "Label": "Primary data sources", "FieldType": "textarea", "HelpText": "Identify systems and sensors (PLC tags, MES, SCADA, quality inspection data, images, operator logs) you will use and where the data currently lives. Note ownership and access constraints.", "Required": true }, { "Key": "data_quality_checklist", "Label": "Data readiness checklist", "FieldType": "checkbox", "HelpText": "Select readiness items that you have validated or can validate within the pilot timeline.", "Options": [ { "Value": "complete", "Label": "Sufficient historical coverage (time span / volume)" }, { "Value": "labeled", "Label": "Labeled or annotated examples for supervised models (if required)" }, { "Value": "consistent", "Label": "Consistent schema & timestamps across sources" }, { "Value": "timely", "Label": "Data latency meets use case needs (near real‑time if required)" }, { "Value": "clean_enough", "Label": "Missing value and outlier profile manageable" }, { "Value": "secure_access", "Label": "Access and privacy/security approvals available" }, { "Value": "not_ready", "Label": "Data not ready yet" } ], "Required": true }, { "Key": "safety_and_governance", "Label": "Safety, regulatory & governance constraints", "FieldType": "textarea", "HelpText": "Describe safety, regulatory, or compliance constraints (e.g., cannot automate safety interlocks, must maintain human authority, data residency requirements). Note required approvals and review cadence.", "Required": true }, { "Key": "stakeholder_approvals", "Label": "Stakeholder approvals likely required", "FieldType": "checkbox", "HelpText": "Tick groups whose approval will be needed before pilot deployment to production or operator use.", "Options": [ { "Value": "ops", "Label": "Operations (plant/line management)" }, { "Value": "safety", "Label": "Safety / EHS" }, { "Value": "quality", "Label": "Quality" }, { "Value": "it", "Label": "IT / OT" }, { "Value": "legal", "Label": "Legal / Compliance" }, { "Value": "procurement", "Label": "Procurement / Vendor management" }, { "Value": "none", "Label": "None / only pilot team" } ], "Required": true }, { "Key": "human_in_loop_model", "Label": "Human‑in‑the‑loop interaction model", "FieldType": "select", "HelpText": "Choose the pattern that best fits how people will interact with AI outputs during the pilot.", "Options": [ { "Value": "decision_support", "Label": "Decision support (recommendations, operator acts)" }, { "Value": "alert_only", "Label": "Alert / flag only (operator investigates)" }, { "Value": "human_override", "Label": "Automated suggestion with human override required" }, { "Value": "closed_loop_control_prohibited", "Label": "Closed‑loop control NOT permitted for safety reasons" }, { "Value": "automated_action_with_guardrails", "Label": "Automated action with strict guardrails & rollback" } ], "Required": true }, { "Key": "minimum_viable_success_criteria", "Label": "Minimum viable success criteria (MVS)", "FieldType": "textarea", "HelpText": "Define clear, testable success criteria for the pilot. Example: 'Model precision > 85% on holdout data and reduces false stoppages by 30% in a 4‑week production trial.' Include acceptance thresholds and measurement method.", "Required": true }, { "Key": "rollback_and_fallback_rules", "Label": "Rollback rules & fallback plan", "FieldType": "textarea", "HelpText": "Specify when and how the pilot will be rolled back (triggers, who authorizes rollback, how to restore previous state). Include data retention, how to stop automated actions, and how to notify stakeholders.", "Required": true }, { "Key": "timeline_and_milestones", "Label": "Pilot timeline & milestones", "FieldType": "textarea", "HelpText": "Provide a short timeline with major milestones (data access, baseline measurement, model prototype, pilot trial, evaluation). Keep it time‑boxed (ideally 4–12 weeks depending on complexity).", "Required": true }, { "Key": "responsibilities_and_lead", "Label": "Team lead and core responsibilities", "FieldType": "textarea", "HelpText": "Name the pilot lead and core team members (ops, data science, IT, quality) and summarize responsibilities (who owns metric measurement, deployment, monitoring, and rollback).", "Required": true }, { "Key": "impact_score", "Label": "Quick impact score (1 low — 5 high)", "FieldType": "scale", "HelpText": "Team‑based ordinal estimate of potential business impact on chosen KPIs.", "Options": { "Min": 1, "Max": 5 }, "Required": true }, { "Key": "feasibility_score", "Label": "Quick feasibility score (1 low — 5 high)", "FieldType": "scale", "HelpText": "Estimate technical and operational feasibility given current data and constraints. Combine data readiness, access, and team capacity in your judgment.", "Options": { "Min": 1, "Max": 5 }, "Required": true }, { "Key": "notes_and_risks", "Label": "Key risks, assumptions & open questions", "FieldType": "textarea", "HelpText": "Capture important assumptions, known risks, and questions that must be resolved for the pilot to succeed.", "Required": false }, { "Key": "priority_suggestion", "Label": "Pilot priority suggestion (manual guidance)", "FieldType": "textarea", "HelpText": "Use the impact + feasibility scores to guide priority. Example rule: high priority if sum ≥ 8, medium if 6–7, low if ≤5. Document the team's rationale for the priority chosen.", "Required": false } ] }