Skip to content
Braintrust → EvalGuard

Datasets and logs. Plus security and guardrails. 

Braintrust does datasets, experiments, scorers, and production logging well. EvalGuard imports both — your log spans and your eval datasets — and adds red-team scans (300+ plugins), an LLM firewall, and a SOC 2 evidence engine on the same data. No sign-up needed to run your first eval.

SOC 2 evidence engineISO 42001 mappedEU AI ActGDPR

Honest positioning

Where Braintrust stops, EvalGuard keeps going

Braintrust is a solid eval and observability tool — datasets, experiments, scorers, and logs done well. EvalGuard overlaps on all of that, then extends into the security, firewall, and compliance work you'd otherwise buy separately.

CapabilityBraintrustEvalGuard
Datasets & experiments YesYes — versioned datasets + run history
Scorers / autoevalsYes — 24 in autoevalsYes — 200+ scorers (LLM-as-judge, pairwise, rubric)
Production logging & tracing YesYes — OTel trace ingest + cost attribution
Prompt playground & versioning YesYes — Prompt IDE + optimizer across 90+ providers
Human review / annotation YesYes — annotation queues with multi-reviewer assignment
AI proxy / gatewayYes — Gateway (the older AI proxy is deprecated)Yes — BYOK gateway, 15 proxied providers, similarity response cache
Red-team & security scansSecurity / moderation judges in autoevals; no attack generatorYes — 300+ attack plugins + 100+ adversarial strategies
LLM firewall / guardrailsNo request-time blocking documented — evals and release gatesYes — real-time input + output firewall
SOC 2 evidence automationNot a documented product featureYes — live evidence engine + audit log
Self-hostingHybrid — data plane in your AWS accountYes — self-host available

Braintrust claims on this page were verified against their source braintrust-sdk-python @ a82dc20e, autoevals @ b0e1055, bt-data-plane @ 2252fbd, plus braintrust.dev/docs raw HTML — on 2026-08-10. Capabilities move; check their current docs before you decide. “SOC 2 evidence automation” describes a product capability, not any vendor’s own certification status.

Migration paths

Braintrust is two things. So the move is two paths.

Braintrust owns your export end-to-end — we never touch your Braintrust account. Convert what you download with the EvalGuard CLI.

A

Observability — bring your log spans

Export your logs from Braintrust (the SDK span.export() / project-logs fetch, or the API), then convert them to neutral-shape spans — model, tokens, scores, and latency all map over.

# Logs/spans → neutral spans (epoch-seconds timing handled for you):
npx @evalguard/cli import:traces --from braintrust braintrust-logs.json --output spans.json
B

Datasets & evals — convert and run

Export a Braintrust dataset, convert it to a runnable EvalGuard config, then run it — keyless on your machine, or on the EvalGuard cloud.

1
Export your dataset from Braintrust
# JS: for await (const row of initDataset("My App", { dataset })) … # Python: initDataset(project, dataset=...).fetch()
2
Convert it to an EvalGuard config
npx @evalguard/cli import:braintrust braintrust-dataset.json -o evalguard.config.json
3
Run it keyless — no account, no API key (echo provider)
npx @evalguard/cli eval:local evalguard.config.json --provider echo

The echo provider runs the whole eval loop with no API key — perfect for validating the config before you spend a token. Swap in --provider openai (with your key) for real model calls, or run on the cloud for shared dashboards and run history:

# Run on the EvalGuard cloud (log in first):
evalguard eval --project <id>

Want a hand with the migration?

If you have a large log history or a dataset with custom scorers, send us your Braintrust export and we'll help you map it and validate the first run. Free.