Module 06 · AI data paths
runnableData Lineage
Show where an AI request was meant to go.
Runnable v0.1 slice available on GitHub — architecture and scope below.
Problem
Once a prompt crosses a gateway, it can be hard to explain which provider, region, model, or internal service handled it. Existing lineage tools cover data estates well, but AI request paths often need a smaller, trace-linked view.
Where it sits
Sits beside Smart Router and records the intended path of AI-bound data. It is an add-on to model traffic, not another LLM or a replacement for a broader data catalog.
Data Lineage is designed to follow requests already passing through Smart Router. Using request and trace IDs, it can show the selected route, referenced provider location, model target, and policy that informed the decision.
This is evidence of routing intent and observed hops, not a blanket residency guarantee. OpenMetadata, Atlan, or another catalog can remain the broader system of record; this module is meant to add the AI request path those systems may not capture directly. Tooling value is the add-on—not a new foundation model.
How it fits
Data Lineage observes the same Smart Router path as Evals: one explains where a request went, while the other records how it scored against evaluation and policy rules. The control plane hub can place both beside the route without turning them into one coupled service.
Capabilities
- Intended request and trace ID correlation with Smart Router
- Route views across local, on-prem, and cloud targets
- Residency-policy evidence attached to routing decisions
- Observed-hop records for AI-bound requests
- References into existing catalog and governance systems
- Focused lineage add-on without introducing another model
Other modules
- Smart RouterModel access gets scattered across provider keys, endpoints, deployment rules, and separate UIs. Even a VS Code assistant and a home chatbot can end up repeating the same integration work.
- Evals & English GovernanceEach model platform has its own eval screens, scoring terms, and policy controls. Teams end up comparing results by hand, while governance rules live in documents that do not map cleanly to provider settings.
- Stage GateAI-assisted projects can jump from an appointment straight into generated code. The missing middle—prerequisites, requirements, decisions, and clear advance criteria—is usually spread across documents and understood by only a few people.
- Shadow VersioningOne project can have its brief in Confluence, approvals in SharePoint, code in Git, and evidence in a testing system. Humans see one piece of work, but the tools expose unrelated histories and technical version language.
- Context EngineAn agent working across 15 repositories can spend more tokens rediscovering the codebase than answering the question. Every tool builds its own partial picture, then throws much of that work away.
- AI-DORAAI coding agents inflate commit counts and PR volume, hiding code churn, review bottlenecks, and rework. Traditional DORA dashboards track deployment frequency without telling you if half the AI-written lines were reverted two weeks later.