Public roadmap
What is live, what is next, and where the agent is heading.
Omnitrace is being built with design partners around one principle: every roadmap item should move a Databricks team closer to verified cost savings or fewer reliability incidents.
50+
detector types live
19
auto-fix paths live
3
autonomy guardrails
Available now
Available for Databricks-focused environments.
FinOps Agent
Finds Databricks waste, quantifies it in dollars, and routes fixes through guardrails.
Reliability Agent
Detects Spark, SQL, table, and cluster drift before repeated failures become platform incidents.
50+ detector types
Coverage across compute waste, SQL warehouse spend, Spark reliability, table hygiene, storage, ownership, and governance.
19 auto-fix paths
Approved low-risk actions can move through the apply-and-verify loop with audit evidence.
Jira workflow
Findings become owner-ready tickets with root cause, estimated impact, and recommended action.
Two-tier verification
Actions are verified with scheduled polling and targeted read-back checks after apply.
Shipping next
The next set of capabilities we are prioritizing with design partners.
Live ROI dashboard
Roll up detected, approved, applied, and verified savings by workspace, team, and strategy.
Richer Spark telemetry
Bring deeper JVM, shuffle, spill, IO, and GC signals into the reliability agent loop.
Policy and approval templates
Preset autonomy policies for FinOps, SRE, platform engineering, and executive review.
Self-serve pilot onboarding
Connect a workspace faster, validate read-only discovery, and generate the first savings thesis.
Planned expansion
Directionally committed areas that become important as customer environments scale.
Multi-platform compute
Extend the same agent loop to EMR, Starburst, Snowflake, and hybrid lakehouse estates.
Spark config auto-tuning
Recommend and verify executor sizing, shuffle tuning, adaptive execution, and query-pattern fixes.
Private deployment
Run inside a customer's VPC for stricter data residency, compliance, and network-control needs.
SSO, RBAC, and enterprise governance
Role-based approval controls, audit views, and enterprise identity integration.
Longer horizon
Bigger platform ideas we expect to shape with early enterprise customers.
Tool marketplace
Bring your own MCP servers, remediation strategies, and environment-specific tools.
Multi-agent collaboration
Specialized investigator, remediator, governance, and cost agents working on the same evidence graph.
Open agent framework
A path for community and customer-contributed detectors, strategies, and verification logic.
Want to shape the roadmap?
Early customers get direct input on prioritization. Tell us what your Databricks team needs before rollout.