Summary
Add first-class Lakemeter estimation for Databricks Agent Bricks, including Knowledge Assistant and Supervisor Agent workloads.
Billing data distinguishes AGENT_BRICKS and SUPERVISOR_AGENT billing origins, but Lakemeter does not currently expose an Agent Bricks workload or calculation model.
Scope
- Verify current public pricing, shared SKUs, billing units, free allowances, and additive underlying services from official Databricks sources.
- Model Knowledge Assistant and Supervisor Agent separately when their billable dimensions differ.
- Identify costs that must be estimated separately, such as Model Serving, Foundation Model API, AI Search, evaluation, or storage.
- Use practical planning inputs that can be reconciled with
system.billing.usage metadata.
- Avoid inventing rates when only a billing origin, rather than a standalone SKU, is published.
Acceptance criteria
- Add appropriate Agent Bricks workload type or types, calculation APIs, and frontend forms.
- Resolve exact cloud, region, tier, workload mode, and SKU pricing without cross-region fallback.
- Persist all new fields in
workload_config; no database schema migration.
- Show transparent monthly usage → DBUs → SKU rate → cost formulas.
- Include AI-assisted estimate creation, cloning, and Excel export.
- Explain additive and excluded underlying service costs in the UI.
- Add calculation, validation, persistence, parity, and export tests.
- Validate assumptions against official documentation and billing system-table fields before release.
Summary
Add first-class Lakemeter estimation for Databricks Agent Bricks, including Knowledge Assistant and Supervisor Agent workloads.
Billing data distinguishes
AGENT_BRICKSandSUPERVISOR_AGENTbilling origins, but Lakemeter does not currently expose an Agent Bricks workload or calculation model.Scope
system.billing.usagemetadata.Acceptance criteria
workload_config; no database schema migration.