What's new in Domino

Domino workspaces provide persistent, governed context across the full AI lifecycle. Coding assistants ship pre-installed with full access to your project, data, and infrastructure.

Why it matters:

• Everything is captured in the audit trail, with full parameter lineage and reproducibility
• Agents see your project files, datasets, and platform resources.
• Three agents, zero setup: Copilot, Claude Code, and Codex are ready to use the moment you open a workspace.

Domino enables teams to securely host and serve large language models (LLMs) within their own infrastructure, deploying them as high-performance GenAI endpoints.

Benefits:

• Host LLMs inside private infrastructure or VPCs with fine-grained access controls
• Reduce operational costs with efficient model storage and runtime mounting
• Maintain full observability into latency, token usage, and system behavior

Domino introduces framework-agnostic instrumentation for agent-based systems, enabling teams to trace and monitor agent interactions from development through production using a single SDK.

Benefits:

• Capture detailed traces including inputs, outputs, tokens, latencies, and tool calls
• Gain full visibility into agent behavior across complex, multi-step workflows
• Automatically associate traces with prompts, models, datasets, and code versions

Teams can visualize, evaluate, and compare agentic systems with both summary and trace-level detail, using shared metrics and complete configuration lineage to support consistent, repeatable evaluation.

Benefits:

• Compare prompts, models, parameters, and agent configurations side by side
• Drill down from aggregate metrics to individual interaction traces
• Evaluate different frameworks and models in the same experiment

Domino closes the gap between experimentation and deployment by enabling teams to deploy agentic systems directly from evaluation results, with the system of record ensuring what is deployed exactly matches what was evaluated.

Benefits:

• Eliminate manual handoffs and brittle deployment scripts
• Reduce risk by deploying fully evaluated, traceable systems
• Maintain complete lineage and auditability across environments

The same instrumentation used during development continues to operate in production, enabling continuous or scheduled monitoring and evaluation of real-world agent behavior, user feedback, and rapid restoration into a workspace for iteration.

Benefits:

• Monitor production performance with real-time traces and dashboards
• Detect regressions through scheduled evaluations
• Rapidly iterate by reopening production systems with one-click for further experimentation