Introducing DigitalOcean Signals and Simulations: see how your agents and models really behave in production, then test the fix before it ships
DigitalOcean launched 'Signals' for automated production traffic triage and 'Simulations' for multi-turn scenario testing within Agent Learning Services. These tools integrate directly with DigitalOcean Managed Agents, Inference Engine, Harness Runtime, and Action Gateway.
Verified State Diff
Impact & Verification Analysis
Developers and AI engineers building and deploying agentic workflows on DigitalOcean Managed Agents and Inference Engine.
It bridges the gap between production observability and development testing, reducing the manual overhead of debugging non-deterministic agent behavior and improving the reliability of LLM-based applications.
Full Fact Overview
DigitalOcean has expanded its Agent Learning Services stack by introducing two observability and testing layers. 'Signals' performs automated triage on production traffic by classifying interaction, execution, and environmental signals, utilizing a trajectory sampling method to identify failures that standard HTTP-level observability misses. 'Simulations' allows developers to convert flagged production sessions into multi-turn evaluation scenarios, enabling pre-deployment validation using the same infrastructure harness. The system is designed to handle high-volume trace data (up to 100 GB/s) by using an inexpensive triage pass before applying LLM-based grading, and it includes built-in PII redaction.