Product Innovation Studio
Outcome
Give the innovation lead a concept-to-gate workspace that turns consumer insight into formulation and claim options, checks each option against ingredient restrictions, allergen rules and label regulations in the target markets, and assembles the stage-gate package for review.
| Portfolio attribute | Value |
|---|---|
| Primary persona | Innovation lead |
| Primary aggregate | InnovationConcept |
| Cosmos DB containers | 2 |
| Grounding data | insights, formulas, claims, regulations, portfolio |
| Development estimate | USD 60-220 monthly dev estimate |
| Runtime options | Python FastAPI and C# ASP.NET Core |
Productivity hypothesis
These measures reflect the workflow, evidence, decision boundary, and industry risk for this use case.
| Impact type | Measure | Conservative internal assumption | Customer measurement |
|---|---|---|---|
| Productivity | Concept development cycle time | 8-15% reduction | (baseline median concept-development minutes - assisted median minutes) / baseline median minutes * 100 |
| Quality or risk | First-pass acceptance | 5-10 percentage-point improvement | assisted percentage accepted without material rework - baseline percentage accepted without material rework |
Important: Conservative internal pilot assumptions only; not public benchmarks, commitments, or expected customer results. Validate against a comparable baseline or matched control cohort with mandatory human review.
Rationale: Innovation teams can combine insights, formulas, claims, regulations, and portfolio context into faster compliant drafts.
Pilot guardrail: Do not generate unsupported product claims or assume formula, regulatory, or manufacturing feasibility.
Customer validation plan
- Baseline the current workflow with representative users and tasks.
- Run the accelerator on the same task mix with human review retained.
- Measure concept development cycle time and first-pass acceptance with the documented formulas.
- Replace internal assumptions with customer-observed results before a business case.
Why Azure Cosmos DB
This AI solution needs an operational data plane, not only a vector index:
- Hybrid vector and full-text retrieval combines semantic recall with exact identifiers, codes, names, and domain language.
- Hierarchical partition keys and metadata predicates keep tenant and domain boundaries in the normal query path.
- Operational JSON, source chunks, workflow and agent state, feedback, and audit records remain together for low-latency reads and governed writes.
- Change feed reacts to source and operational updates without polling.
- Autoscale throughput, configurable consistency, and multi-region distribution support variable demand and enterprise availability.
Domain workflow

| Stage | Actor | Input → output | Approval |
|---|---|---|---|
| 1. Insight-to-proposition framing | Innovation lead with product innovation studio | consumerInsightRef → Concept | Policy controlled |
| 2. Formulation option generation | Product innovation studio with R and D scientist | Concept → FormulaCandidate | Policy controlled |
| 3. Market restriction and allergen screening | Regulatory screening service | FormulaCandidate → restrictionFindings | Policy controlled |
| 4. Claim substantiation review | Regulatory affairs reviewer | ClaimCandidate → approvedClaimSet | Human required |
| 5. Stage-gate decision | Gate body | GateAssessment → gateDecision | Human required |
Technical architecture

Cosmos DB data model
| Container | Hierarchical partition key | Documents | Search paths |
|---|---|---|---|
innovationConcepts | /tenantId, /conceptId | Concept, FormulaCandidate, ClaimCandidate, GateAssessment | /propositionVectorEmbedding, /claimVectorEmbedding, /propositionVector, /claimVector |
productRegulationLibrary | /tenantId, /marketCode | RegulatoryChunk | /chunkVectorEmbedding, /chunkVector |
Query contract
| Query | Container | Scope | Purpose |
|---|---|---|---|
| ConceptsAwaitingGate | innovationConcepts | cross-partition-bounded | List concepts whose assessment is assembled and awaiting a gate decision. Intentional cross-partition stage-gate queue: bounded by the tenantId prefix, one brand, one gate number and one gate owner. |
| BlockedFormulaCandidates | innovationConcepts | single-partition | Show candidate formulations carrying unresolved restriction findings in any target market. |
| MarketRuleRetrieval | productRegulationLibrary | single-partition | Retrieve ingredient and labeling rules in force for a target market and rule category. |
| ClaimsBlockedByMarket | innovationConcepts | single-partition | Identify claims that cannot run in one or more target markets before artwork is commissioned. |
Python and C# services route each document type to its declared container and hierarchical partition key. ETag-based writes protect mutable aggregates, while append-only records and bounded cross-partition queries preserve scale.
Azure deployment

Clone the source repository, authenticate with Azure, and run:
Set-Location accelerators/consumer-packaged-goods/product-innovation-studio
az login
./Deploy.ps1 -Runtime python -Location eastus2
Change python to csharp to deploy the .NET implementation.
Explore the implementation
- Launch signed-in demo
- Implementation article
- Domain data model
- Workflow and API contract
- Python application
- C# application
- Domain container Bicep
- Platform Bicep
- Cost and configuration
- Source repository