DBOS vs Hatchet (2026)

Native integrations with Pydantic AI, LlamaIndex, the OpenAI Agents SDK, Google ADK and the Vercel AI SDK make the customer's agent framework durable instead of supplying the tools. They wrap the agent loop as a workflow and each model request, tool call and MCP communication as a checkpointed step, and any step can wrap an outside API call with retries. The tools, credentials and connectors belong to the framework and the customer's code, and DBOS has no connector catalog or framework of its own for authenticated actions.

Sourcedocs.dbos.dev/integrations/pydantic-airead 2026-09-23

The customer's agents and tools run here, but Hatchet does not supply them. Cookbooks show agents built with the Claude Agent SDK and the OpenAI Agents SDK running as Hatchet tasks, and any task can call an outside API. There is no connector catalog, custom tool route or framework for authenticated actions of Hatchet's own, so the tools and credentials are the customer's code. Webhook integrations with GitHub, Stripe and Slack start runs.

Sourcedocs.hatchet.run/cookbooks/hatchet-claude-agent-sdk-trusted-envread 2026-09-23

Durable orchestration is the product. Workflows and steps are written as ordinary code in Python, TypeScript, Java or Go, with automatic step retries, durable queues carrying concurrency and rate limits, child workflows, durable sleeps and waits, messages and events between workflows, and agent loops (including subagents through the Vercel AI SDK integration) run as workflows.

Sourcedocs.dbos.dev/ai/ai-quickstartread 2026-09-23

Tasks compose into DAG workflows and durable tasks that can spawn child tasks, so Hatchet works as a workflow engine. They carry retry policies, timeouts, cancellation, durable sleeps and event waits, along with concurrency control, rate limits, priority, batch tasks and bulk triggering. Workers run them in Python, TypeScript, Go and Ruby.

Sourcedocs.hatchet.run/v1/directed-acyclic-graphsread 2026-09-23

Work starts on its own. Scheduled workflows run exactly once per interval on cron schedules, which can be created at run time and are stored in the database, and durable queues start enqueued workflows as capacity allows.

Sourcedocs.dbos.dev/golang/tutorials/scheduled-workflowsread 2026-09-23

Work wakes on its own through cron runs, scheduled runs, events that trigger workflows, and webhook endpoints that start runs from GitHub, Stripe, Slack or any sender, and one service can trigger work in another.

Sourcedocs.hatchet.run/v1/webhooksread 2026-09-23

There is no retrieval structure DBOS maintains over the customer's corpus. Workflows are checkpointed to the customer's own Postgres, and the Vercel AI SDK integration makes embedding calls durable steps, but the index and retrieval belong to the customer's code.

Sourcedocs.dbos.dev/examplesread 2026-09-23

There is no retrieval structure that Hatchet maintains over the customer's corpus. Its cookbooks run document pipelines as tasks, but the parsing, indexing and retrieval are the customer's own code.

Sourcedocs.hatchet.run/cookbooks/pdf-pipelineread 2026-09-23

State is kept inside one run. Every workflow and step is checkpointed to Postgres, so a workflow that crashes resumes from its last completed step and the agent's code carries on from the restored outputs. The scope is one run, so a checkpoint lets the same run finish and carries nothing into the next. Step outputs sit in queryable tables, and Conductor shows the history. There is no state store that persists across runs for each user, conversation or entity as agent memory.

Sourcedocs.dbos.dev/production/retentionread 2026-09-23

State lasts within one run. Hatchet is durable by default and stores every task, DAG, event and agent invocation in a durable event log, so a durable task replays its completed steps and carries on after a failure. But the checkpointed state only lets the same run finish and carries nothing into the next run. Sticky worker assignment keeps local memory on one worker for a workflow's tasks, which is placement, not a store. There is no state store that persists across runs, keyed per user, conversation or entity, for agent memory.

Sourcedocs.hatchet.run/v1/durable-executionread 2026-09-23

Human input is handled with a durable wait in the customer's code. In the human in the loop guide, an agent calls DBOS.recv to wait hours or days for a notification with a deadline, checkpointed so the wait survives restarts and upgrades, while DBOS.send delivers the person's response and workflow events publish the agent's status. That is a mechanism the customer's code calls, but it is general durable messaging applied to approval, with no approval step, approver identity, review surface or policy guardrail of DBOS's own, and the gate's logic is the customer's code.

Sourcedocs.dbos.dev/ai/hitlread 2026-09-23

Holding for a person runs on durable waits that the customer's code calls. A durable event wait suspends a task until a matching user event arrives, filtered with CEL expressions, or until a timeout passes, and workflows can be paused and resumed. This is a general wait put to use for approval. There is no approval step, approver identity, review screen or policy guardrail of Hatchet's own.

Sourcedocs.hatchet.run/v1/durable-event-waitsread 2026-09-23

DBOS states SOC 2 compliance, audited regularly, along with GDPR and CCPA, with a HIPAA BAA available and SOC 2 and HIPAA support on Teams. Teams adds role based access control and Enterprise adds SSO and SAML, and Conductor's organization management supports RBAC and any OAuth compatible SSO. DBOS also states that Conductor is off the execution path and never sees workflow data.

Sourcedbos.dev/dbos-pricingread 2026-09-23

SOC 2 Type II applies from the free Developer plan, and HIPAA with audit logs from the Scale plan. Tenants separate environments, each with its own users. Hatchet Cloud signs users in with Google or GitHub OAuth, enterprises can set up custom SSO, and Enterprise includes SSO and audit logging.

Sourcehatchet.run/pricingread 2026-09-23

Each workflow is recorded step by step. Every step's input and output is checkpointed, and the Conductor console shows workflow state, history and traces, monitors queues in real time and alerts on failures (custom alerting on Teams), with managed retention policies setting how much history each application keeps and for how long. The library emits OpenTelemetry traces and logs, OpenMetrics is supported on Teams, and an MCP server lets the customer's own agents inspect workflows.

Sourcedocs.dbos.dev/production/conductorread 2026-09-23

Every task, DAG, event and agent invocation is recorded in a durable event log shown in the dashboard, with real time monitoring, alerting and task logs. Traces export through OpenTelemetry, Prometheus metrics are exposed, audit logs are kept from the Scale plan, and data retention is set by plan, at 3 days on Team and 7 on Scale.

Sourcedocs.hatchet.run/v1/opentelemetryread 2026-09-23

The customer decides where DBOS runs. The open source library runs inside the customer's application on any Linux, macOS or Windows host, on premises or in any cloud, against the customer's own Postgres, and Conductor can be self hosted (with Kubernetes guides) and run in air gapped environments on Enterprise.

Sourcedocs.dbos.dev/production/hosting-conductorread 2026-09-23

Customers can self host in their own infrastructure on Docker Compose or Kubernetes under an MIT license, and there is an embedded mode. Hatchet Cloud is the managed service, and Enterprise offers custom deployment and bring your own cloud.

Sourcegithub.com/hatchet-dev/hatchetread 2026-09-23

Quickstarts and an examples library ship in each language, including AI examples such as an agent inbox and a customer service agent. These are examples a developer builds from, not a catalog of agents a customer adopts.

Sourcedocs.dbos.dev/examplesread 2026-09-23

Starting points come as cookbooks for webhook handlers for GitHub, Stripe and Slack, document pipelines, email flows and agents built on the Claude Agent SDK and OpenAI Agents SDK, plus skills that coding agents can install. These are examples a developer builds from, not a catalog of agents a customer adopts.

Sourcedocs.hatchet.run/cookbooksread 2026-09-23

No model runs inside DBOS for a customer to choose. The model lives in the customer's own framework (Pydantic AI, the OpenAI Agents SDK, Google ADK, the Vercel AI SDK), and DBOS checkpoints each model request as a step whatever the model is. Being model agnostic because the model belongs to someone else is different from offering a choice of model.

Sourcedocs.dbos.dev/integrations/vercel-airead 2026-09-23

There is no model in Hatchet for the customer to choose. The model lives in the customer's own agent code, and the cookbooks use the OpenAI Agents SDK and the Claude Agent SDK, while Hatchet schedules and records the tasks whatever the model is. It works with any model only because the model belongs to someone else, and it offers no model selection of its own.

Sourcedocs.hatchet.run/cookbooks/hatchet-openai-agents-sdk-trusted-envread 2026-09-23

Outside systems can call DBOS through open source SDKs for Python, TypeScript, Java and Go, a Conductor API with a published OpenAPI specification for managing workflows in code, the dbosctl command line tool, and an MCP server for inspecting workflows.

Sourcedocs.dbos.dev/production/conductor-apiread 2026-09-23

Official SDKs for Python, TypeScript, Go and Ruby make the platform callable from outside, alongside a REST and gRPC API for triggering and inspecting runs and the hatchet CLI. A local MCP server bundled with the CLI lets coding agents trigger runs, inspect status and events, and list workers.

Sourcedocs.hatchet.run/v1/using-coding-agentsread 2026-09-23

Debugging after a failure is supported. The debugging guide uses the checkpointed history to find where an agent went wrong and workflow fork to restart it from a chosen completed step, reproducing the state up to that point deterministically, in code or from the Conductor UI. That is debugging, not testing. There is no evaluation harness, scored test cases or quality gate of DBOS's own, and the example evaluation step is the customer's own LLM call.

Sourcedocs.dbos.dev/ai/debuggingread 2026-09-23

When a run goes wrong, it can be replayed from the durable event log, retried or canceled in bulk, and inspected by a coding agent through the MCP server. That helps with debugging, not testing. There is no evaluation harness, scored test cases or quality gate of Hatchet's own.

Sourcedocs.hatchet.run/v1/durable-executionread 2026-09-23

DBOS operates no browser or desktop for the agent and has no browser automation environment of its own. It makes the customer's code durable wherever it runs.

Sourcedocs.dbos.dev/ai/ai-quickstartread 2026-09-23

The agent operates no browser or desktop, and Hatchet has no browser automation environment of its own. It schedules and runs the customer's tasks on the customer's workers.

Sourcedocs.hatchet.run/v1/using-coding-agentsread 2026-09-23