EXEMPLARExemplar is part of the NVIDIA Inception program.
NVIDIA InceptionExemplar is part of the Google for Startups program.
Google for StartupsMarshal · Regression gates for agents
Evals
Evaluation suites for prompts, skills, and agent workflows—regression gates before changes reach production traffic
Marshal · Evals
How Evals works
Evaluation suites for prompts, skills, and agent workflows—regression gates before changes reach production traffic
EVALS
ingest · score · review
Sessions land in the ingest store (sky rails); scores and pass rates flow out for review (orange rails).
Sessions land in the ingest store (sky rails); scores and pass rates flow out for review (orange rails).
Why it matters
Built for production agents—not prompt playgrounds
How this harness capability fits the Exemplar platform: one identity, one policy layer, and one audit trail across console, MCP, and SDK.
Why Exemplar
Shipping prompt or skill changes without evals is deploying code without tests—especially risky when agents can touch production.
Exemplar evals sit in the harness control plane, not a separate lab notebook disconnected from live policy and tools.
What Exemplar delivers
Managed eval datasets and graders for operational agent behaviors—not just chat quality.
CI-style gates: failed evals block prompt rollouts and orchestration publishes.
How teams use it
Define golden scenarios from past incidents and Day 2 Ops runbooks; run on every prompt or skill change.
Compare scores across models and disclosure strategies before shifting traffic.
Capability checklist
Developer guide
Official documentation on docs.exemplar.dev for this capability.
Open developer guide (opens in a new tab)Get started
Put Evals to work with the rest of Marshal—or talk to us about your agent stack.
From the blog
Related posts on exemplar.dev.
- The Harness Engineering Checklist
15 things to put in place before trusting AI-generated code in production — organised by phase: foundation, enforcement, task design, and maintenance. The checklist most teams wish they had before they started.
- AI Agent Governance: How to Control AI Agents Running in Production
AI agent governance is the set of policies, controls, and audit mechanisms that determine what AI agents can do, when they need human approval, and how their actions are logged. The five pillars, how governance differs from the harness, and why it matters for compliance.