Agentic systems orchestrate large language models (LLMs) with tools, APIs, and multi-step reasoning to accomplish complex tasks. These systems rely on LLM endpoints to process requests and generate responses. Register and deploy LLMs covers setting up the infrastructure your agents will use.
Experimenting with agentic systems is not the same as training traditional ML models. Instead of starting from scratch, you assemble systems that mix LLMs, prompts, tools, and agents. These systems are powerful but also complex, dynamic, and challenging to evaluate.
Domino provides a complete workflow for developing, experimenting with, deploying, and monitoring agentic systems:
- Develop
-
Instrument your code to capture traces during execution. Traces record every step your agentic system takes, including downstream calls, so you can debug, analyze, and explain results.
- Experiment
-
Attach evaluations to traces to assess quality. Compare different configurations to see how prompts, agents, or models affect outcomes. Export results for analysis and sharing.
- Deploy
-
Launch your best configuration to production. Domino continues collecting traces from real user interactions automatically.
- Monitor
-
Track performance with production traces. Run evaluations continuously and iterate when you identify issues or opportunities for improvement.
These capabilities help you track experiments, evaluate results, and iterate with confidence. Traces, evaluations, and comparisons live inside your project, making it easy for colleagues to review, share, and extend your work.