Native integration
DSPy
Run predictions using self-hosted DSPy programs
Integrate with your self-hosted DSPy programs for LLM-powered predictions. Supports Predict, Chain of Thought, and ReAct agents. DSPy is the framework for programming—not prompting—language models.
What teams can do
Move from signal to action.
Combine DSPy with reasoning, approvals, data, and every other SteelEngine block. The result is one governed workflow rather than another brittle handoff.
Combine deterministic actions with agent reasoning
Ship with controls, approvals, and a full audit trail
Connection path
From account to production in three moves.
SteelEngine keeps credentials encrypted and the workflow visible. Add DSPy, choose the capabilities an agent can use, and ship with the controls your team requires.
Authorize DSPy
Connect with api key and keep credentials out of workflow logic.
Choose actions and triggers
Use fixed workflow steps, expose selected actions to an agent, or combine both.
Add production guardrails
Apply approvals, permissions, retries, schedules, and an audit trail before deployment.
Supported capabilities
3 ways to work with DSPy.
Capabilities are available as workflow blocks and can be selectively exposed to agents. Exact configuration and permission requirements are documented for each action.
Predict
Run a prediction using a self-hosted DSPy program endpoint
Chain of Thought
Run a Chain of Thought prediction using a self-hosted DSPy ChainOfThought program endpoint
ReAct Agent
Run a ReAct agent using a self-hosted DSPy ReAct program endpoint for multi-step reasoning and action
Build with DSPy
Put the connection to work.
Start with a real operational workflow. SteelEngine’s team can help map the systems, controls, and deployment path.