LoginSign Up
← Back to all sparks
C

Comet

AI-ASSISTANTS
Velocity5.0

ML experiment tracking and LLM observability platform, including Opik for evaluating LLM apps.

www.comet.com ↗

Comet's Opik pushes eval and observability toward standardized, portable agent workflows.

ai-evaluationobservabilityagentsinteroperabilityeval-driven-development
◆Current state
Comet is centering its Opik product on AI evaluation and agent observability — test suites, tracing, and evaluation-driven development for teams shipping agents to production. Recent moves include an integration with Oracle's Open Agent Specification and automated eval workflows.
◆Where it's heading
The direction is toward measurable, portable agent development: build-once-run-anywhere via open specs, automated dataset and metric evaluation, and deep tracing to debug multi-step agent failures. Comet is planting itself as the eval/observability layer for the agentic stack.
◆Prediction
Expect more eval automation and interoperability work — additional framework integrations and tooling that treats every agent change as a measured experiment.

◆Recent moves

  1. 4d ago

    How Evaluation-Driven Development (EDD) Works

    An explainer on evaluation-driven development — treating each agent change as a measured before/after experiment; educational content aligned to Opik but not a release.

    View source ↗
  2. 6d ago

    Opik + Oracle Agent Specification: Build Once, Run Anywhere

    Opik integrates with Oracle's Open Agent Specification, adding portability so agents built to the spec can be evaluated and shipped across frameworks — a real interoperability move, framed by Comet as a partnership.

    View source ↗
  3. 11d ago

    AI Evaluation Simplified: Automate Dataset & Metric Eval Workflows with Test Suites

    Opik adds Test Suites to automate dataset and metric evaluation workflows, reducing the hand-built reference-dataset burden — a concrete new eval capability.

    View source ↗
  4. 11d ago

    Advanced Claude Code Cost Tracking: How to Save 30% on Token Spend

    A how-to on cutting Claude Code token spend; content marketing tied to observability, not a product change.

    View source ↗
  5. 19d ago

    Understanding Your Claude Code Spend: What’s Actually Driving the Cost

    An analysis of what drives Claude Code costs; educational content with no product signal.

    View source ↗
  6. 1mo ago

    Agent Tracing and Observability: Log & Debug Complex AI Systems

    An explainer on agent tracing and observability for debugging AI systems; category education, not a release.

    View source ↗