AgentOps

AgentOps is an observability and debugging platform for AI agents and applications built on large language models. It instruments an agent run so that every model call, tool invocation and interaction between agents is recorded, then presents the whole trace visually. Installation is a single Python package, and the SDK is agent agnostic, with native integrations for OpenAI, CrewAI, Autogen and a long list of other frameworks and model providers. The platform is for engineers who need to understand why an agent behaved the way it did in a run that has already finished. Time travel debugging rewinds and replays a session with point in time precision, so a failure can be reproduced without rerunning the whole workflow. A full data trail keeps logs, errors and records of prompt injection attempts from prototype through production. Cost control is the second focus: token counts are tracked for every agent, spend is monitored against current model prices, and saved completions can be reused to fine tune specialized models more cheaply. The Basic plan is $0 per month and covers up to 5,000 events, with the agent agnostic SDK, model cost tracking and replay analytics. Pro starts at $40 per month on a pay as you go basis with a pricing calculator, and adds an unlimited event limit, unlimited log retention, session and event export, role based permissioning and dedicated Slack and email support. Enterprise is quoted and adds a service level agreement, Slack Connect, custom single sign on, on premise deployment, self hosting on AWS, GCP or Azure, custom data retention and compliance coverage spanning SOC 2, HIPAA and the NIST AI risk management framework. Paid help with building and scaling agents is offered separately.