Progressive Workflow Assessment Seed Handoff
Six onboarding answers can seed AI Fluency hints. They cannot become a workflow fingerprint.
The live AINA Platform six-question onboarding funnel is now represented inside the data engine room as a strict PlatformAssessmentSeed. It can seed weak AI Fluency hints, but it cannot create a workflow fingerprint. Workflow-specific fields stay unknown until stronger evidence exists.
Six MCQ answers: seniority, tool use, AI readiness, mindset, learning goal, and tool variant.
Assessment-only observations, source refs, gap reasons, and proxy quality flags.
Workflow name, frequency, pain, time, stakeholders, output type, constraints, risk, checkpoints, and AI boundaries.
What Changed
The schema layer now has strict platform assessment contracts: PlatformAssessmentSeed, PlatformAssessmentSignals, and PlatformAssessmentAnswerRef. It also has ProgressiveWorkflowEvidenceSummary and ProgressiveWorkflowEvidenceField, so missing workflow evidence is visible instead of silently inferred.
The adapter emits CapabilityObservation(signal_type="assessment") records only. The preliminary capability map keeps all five AI Fluency layers, but its overall score is 0.0 and its band is unknown.
Proof Artifacts
| Artifact | Path |
|---|---|
| Validation JSON | artifacts/validation/progressive_workflow_assessment_seed_v1.json |
| QA JSONL | artifacts/validation/progressive_workflow_assessment_seed_v1.jsonl |
| Markdown report | artifacts/reports/progressive_workflow_assessment_seed_v1.md |
| HTML report | artifacts/reports/progressive_workflow_assessment_seed_v1.html |
| Check | Result |
|---|---|
| Six-question seed valid | pass |
| Capability map keeps exactly five layers | pass |
| Workflow fingerprint not created | pass |
| Compiler workflow fingerprint blocked | pass |
| Workflow-specific fields remain unknown | pass |
| No workflow-specific values inferred from onboarding | pass |
| Proxy observations are not fluency proof | pass |
No new human_review fields in this lane | pass |
Validation
cd /srv/aina/aina-data-engine-room uv run pytest tests/test_ai_fluency_capability_map.py tests/test_progressive_workflow_assessment_seed.py -q uv run ruff check src/aina_data_engine/schemas.py src/aina_data_engine/ai_fluency_capability.py src/aina_data_engine/cli.py tests/test_progressive_workflow_assessment_seed.py tests/test_ai_fluency_capability_map.py uv run aina-data-engine --root /srv/aina/aina-data-engine-room progressive-workflow-assessment-seed uv run aina-data-engine --root /srv/aina/aina-data-engine-room ain-506-p0-gate uv run aina-data-engine --root /srv/aina/aina-data-engine-room ain-510-retrieval-promotion-gate uv run aina-data-engine --root /srv/aina/aina-data-engine-room validate
Focused tests passed with 7 passed. Ruff passed for touched Python and test files. The seed command passed with blocked_workflow_field_count=12 and workflow_fingerprint_created=false. AIN-506 passed. AIN-510 is valid but not promoted; the remaining expected blockers are sensitive mismatch fixture coverage and runtime rollback proof. Full validation passed.
Important Boundary
This slice did not touch /srv/repos/aina-platform. It mirrors the current platform diagnostic shape from read-only source inspection and keeps platform bridge/export as a later slice. It also did not embed learner-specific fingerprint data or promote any public runtime authority.
Older validation lanes still contain legacy review language, but this new lane introduces no human_review fields or gates. The new receipt uses assessment, proxy, quality, and workflow-evidence language only.
Resume Commands
cd /srv/aina/aina-data-engine-room
git status --short --branch
jq '{schema_version, valid, status, metrics, checks, scope}' artifacts/validation/progressive_workflow_assessment_seed_v1.json
uv run aina-data-engine --root /srv/aina/aina-data-engine-room progressive-workflow-assessment-seed
uv run aina-data-engine --root /srv/aina/aina-data-engine-room validate
Next Best Work
The next slice should build the platform bridge/export contract only after comparing the live six-question payload from /srv/repos/aina-platform against this engine-room seed. After that, join the seed with role/title/source-intelligence evidence to move fields from unknown_not_observed to source_intelligence_seeded, learner_reported, or learner_confirmed one field at a time.
Use the seed receipt as the boundary: onboarding can orient the learner, but workflow truth must be earned from stronger evidence.