Use case · Agent evaluation

Catch where your agent breaks - before your users do.

Agents that ace standard benchmarks still fail on the domain-specific, incident-aware questions your users actually ask. Rockfish builds time-series eval suites - grounded in your data - that show exactly where yours falls short.

agentfuel · eval runscoring
What's the highest count of records across all regions?
the analytics agent answered 783
Without AgentFuel
783
✓ looks correct

No ground truth to check against. Ships silently.

With AgentFuel
783
✗ wrong · true 398

Graded against the real number. Caught pre-production.

Same answer, opposite outcome. Ground truth is the difference.

40 of 152 answers in this run were wrong - invisible without it.

The problem

They pass the benchmarks. Then fail your users' real questions.

The problem usually isn't the agent - it's the eval. Generic benchmarks under-test the patterns and questions that actually matter in production. Two gaps cause most of it:

The data gap

Generic evals use flat or made-up data.

Real observability, telecom, and security data has time-aware structure - correlated signals, incident windows, anomaly spikes, and event sequences that unfold over time. An agent tested on generic data has never been tested on yours.

The query gap

Lookups and averages aren't real questions.

Production users ask incident-aware, context-specific questions. An agent that can't answer them isn't production-ready - but a generic benchmark would never catch it.

"Why did latency spike after the 2:15 PM deployment?"
Agents scoring 73% on benchmark-style tasks dropped to 10-20% on domain-specific, incident-focused queries.

The gap between "passes benchmarks" and "ready for your users" is wide - and invisible until you test for it. Read the AgentFuel paper on arXiv →

How it works

From your schema to a bespoke eval suite - in three steps.

AgentFuel turns your schema and domain context into an expressive eval suite, grounded in the real patterns your agent will hit in production.

1

Bring your schema + context

  • Schema or sample data
  • The agent to be tested
  • Your business context
2

Rockfish builds the suite

  • Time-series dataset generated to your domain
  • Scenario injection - incidents, drifts, edge cases
  • Question & answer pairs with ground truth
3

Your custom eval suite

  • Labeled time-series datasets
  • Q&A eval pairs
  • An agent scorecard you can act on
What you get

Ship on evidence, not benchmark optimism.

Find the breaks first

See where your agent fails on time-series and incident-aware questions before a user reports the wrong answer that costs you trust.

Improve continuously

Track improvements and regressions release over release with grounded, domain-specific scorecards and reproducible, versioned benchmarks.

Deploy with confidence

Generic benchmarks show what's possible. Bespoke evals show whether your agent is actually ready for your domain, your data, and your users.

Passing benchmarks isn't the same as being ready.

Find out where your agent actually stands - with an eval suite built for your domain, your data, and your users' real questions.