Observability & traceability
Commercial AI engineering platform that captures traces of agent runs including inputs, outputs, tools and costs, groups them into sessions, visualises agent trajectories, and runs online and offline evaluations with human annotation. Arize also maintains Phoenix, a local-first open-source tracing and evaluation package.
hybrid · generally available · Research snapshot 2026-09-06
Visit the official product source ↗Observability & traceability · Evaluation & testing
Useful conversation with: ML platform lead, AI engineer.
Show me an agent trajectory visualisation for a failed multi-step run and the trace-level evaluation scores attached to it.
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
Instrumentation records each agent run as traces and spans capturing inputs, outputs, tools and costs, with pre-built integrations for 30+ providers and frameworks.
Limit: Docs overview does not describe multi-agent handoff semantics or retention of captured spans.
Documented by provider
Documented evaluation features include continuous evaluator labelling of traces, LLM-as-a-judge, agent-as-a-judge and code evaluators, dataset-based experiments, human annotation of spans, and evaluator alignment against human judgment.
Limit: Evaluator accuracy is not established by the vendor pages.
Documented by provider
Pricing documents session support, multi-modal tracing for image, voice and PDF, agent trajectory path and graph visualisations, trace evals for agent trajectories, session evals for multi-turn conversations, and metered trace-span and ingestion limits per plan.
Limit: Plan feature lists do not state audit logging, retention configuration or regional hosting options.
Documented by provider
Arize maintains Phoenix, described on the pricing page as an open-source, local-first platform for tracing, evaluation, experimentation and prompt iteration.
Limit: Phoenix licensing and feature parity with AX were not verified from a repository page in this session.
Listing does not imply partnership, supplier status, a working DutyGraph integration, or a compliance certification.
Suggest a correction