{
  "version": "0.1.0",
  "generatedAt": "2026-05-16T00:00:00+05:30",
  "purpose": "Client-ready intake schema for Edxperimental Labs consulting diagnostics and benchmark sprints.",
  "contacts": [
    {
      "name": "Sanjay Prasad",
      "email": "sanjay@edxperimentallabs.com",
      "role": "Benchmark design and technical delivery"
    },
    {
      "name": "Saujas",
      "email": "saujas@edxperimentallabs.com",
      "role": "Sales engineering and client solutions"
    }
  ],
  "sections": [
    {
      "title": "Workflow",
      "prompt": "Describe one workflow, the owner, user journey, current process, monthly volume, and what a wrong answer costs.",
      "fields": [
        "notes",
        "files",
        "openQuestions"
      ]
    },
    {
      "title": "Evidence",
      "prompt": "Attach representative documents, tickets, prompts, screenshots, policies, expected outputs, and known failure examples.",
      "fields": [
        "notes",
        "files",
        "openQuestions"
      ]
    },
    {
      "title": "Constraints",
      "prompt": "List latency target, budget boundary, data sensitivity, approved vendors, deployment environment, and compliance requirements.",
      "fields": [
        "notes",
        "files",
        "openQuestions"
      ]
    },
    {
      "title": "Decision",
      "prompt": "State the decision this benchmark must unlock: buy, build, switch, ship, pause, redesign, or run a deeper pilot.",
      "fields": [
        "notes",
        "files",
        "openQuestions"
      ]
    },
    {
      "title": "Candidates",
      "prompt": "List model/provider/agent routes to compare and the baseline process they must beat.",
      "fields": [
        "notes",
        "files",
        "openQuestions"
      ]
    },
    {
      "title": "Acceptance",
      "prompt": "Write the minimum evidence required before the team would trust the workflow in production.",
      "fields": [
        "notes",
        "files",
        "openQuestions"
      ]
    }
  ],
  "proposalMilestones": [
    {
      "time": "Day 0",
      "milestone": "Discovery",
      "output": "Confirm workflow, success metric, risk boundary, data packet, and decision deadline."
    },
    {
      "time": "Day 1",
      "milestone": "Task packet",
      "output": "Write prompts, expected outputs, scoring rubric, holdout policy, and artifact schema."
    },
    {
      "time": "Days 2-4",
      "milestone": "Runs and review",
      "output": "Run candidate systems, capture traces, review failures, and calculate cost/latency envelope."
    },
    {
      "time": "Day 5",
      "milestone": "Decision memo",
      "output": "Deliver ranked recommendation, risk register, next tests, fallback plan, and production-readiness note."
    }
  ],
  "salesEngineeringHandoff": [
    {
      "stage": "Lead qualification",
      "owner": "Saujas",
      "artifact": "Discovery note",
      "evidence": "Workflow, buyer decision, urgency, data sensitivity, budget signal, and stakeholder map."
    },
    {
      "stage": "Technical scoping",
      "owner": "Sanjay Prasad",
      "artifact": "Benchmark scope",
      "evidence": "Task families, candidate systems, scoring rubric, holdout plan, and replay requirements."
    },
    {
      "stage": "Sprint proposal",
      "owner": "Saujas",
      "artifact": "Proposal memo",
      "evidence": "Timeline, deliverables, access requirements, commercial assumptions, and acceptance gates."
    },
    {
      "stage": "Delivery review",
      "owner": "Sanjay Prasad and Saujas",
      "artifact": "Decision packet",
      "evidence": "Run table, trace links, risk register, deployment recommendation, and next sprint."
    }
  ],
  "engagementReadinessGates": [
    {
      "gate": "Workflow specificity",
      "proof": "One workflow with owner, inputs, outputs, volume, and failure cost."
    },
    {
      "gate": "Evidence access",
      "proof": "Representative prompts, documents, tickets, traces, screenshots, or policies are available."
    },
    {
      "gate": "Decision deadline",
      "proof": "The buyer knows whether the sprint must answer buy, build, switch, ship, pause, or redesign."
    },
    {
      "gate": "Candidate systems",
      "proof": "At least two model/provider/agent routes and one baseline process are named."
    },
    {
      "gate": "Review owner",
      "proof": "A human reviewer can judge correctness, partial credit, and unacceptable failures."
    }
  ],
  "deliveryArtifacts": [
    {
      "title": "Workflow risk map",
      "purpose": "Shows where the current process fails, what AI could improve, and what should remain human-reviewed."
    },
    {
      "title": "Benchmark task packet",
      "purpose": "Defines inputs, expected outputs, rubrics, holdouts, and evidence requirements before any model runs."
    },
    {
      "title": "Run and trace ledger",
      "purpose": "Captures model ids, prompts, artifacts, latency, cost, reviewer notes, and failure classes."
    },
    {
      "title": "Deployment decision memo",
      "purpose": "Turns evidence into a recommendation, fallback route, monitoring plan, and next-sprint backlog."
    }
  ]
}