Evaluation & testing
Braintrust
Developer platform for scoring AI application output: teams log traces from their app, curate datasets from human feedback, run experiments against scorers, then watch production logs. Documentation covers project-level access control and a customer-controlled data plane, but audit logging and retention behaviour are not described.
commercial · generally available · Research snapshot 2026-09-06
Visit the official product source ↗Where it fits
Evaluation & testing · Observability & traceability
Useful conversation with: AI engineering lead, Head of ML platform.
Ask for a demonstration
Show me an experiment run comparing two prompt versions against a dataset built from human feedback, and who in my organization can view the underlying logs.
Capabilities and evidence
Support labels reflect the supplied research. Documentation and vendor claims are not independent product tests. “Not established” means the researcher did not find support; it does not prove a capability is absent.
Documented by provider
Braintrust documents capturing traces from an AI application, analyzing logs, adding human feedback to build datasets, and running evals and experiments against them.
Limit: The quickstart does not enumerate span types (LLM call, tool call, multi-agent handoff) or describe retention of logged data.
Source s1
Documented by provider
A self-hosted deployment option lets the customer run the data plane in their own AWS, GCP or Azure account using their own IAM policies, KMS keys and audit trails, while Braintrust hosts the control plane, UI and authentication.
Limit: The audit trails referenced are the customer's own cloud-provider trails, not an in-product immutable audit log of evaluation activity.
Source s2
Documented by provider
Access to organization, project and infrastructure settings is governed by permission groups, with the infra dashboard limited to organization owners and members holding the Manage settings permission.
Limit: No documented evidence for SSO enforcement, SCIM provisioning or per-record retention policies.
Source s1 · Source s2
Limitations to discuss
- Primarily a developer quality-engineering tool; no documented immutable audit record, retention policy or policy-gating control for governance reporting.
- Vendor now markets the product as an 'active observability platform', so the evaluation/observability boundary is blurred.
Sources
- Get started with Braintrust · Braintrust · official docs
Access date reported by researcher: 2026-09-06 - Self-hosting Braintrust · Braintrust · official docs
Access date reported by researcher: 2026-09-06
Listing does not imply partnership, supplier status, a working DutyGraph integration, or a compliance certification.
Suggest a correction