From the course: Agentic AI: Build Your First Agentic AI System
Join today to access over 25,600 courses taught by industry experts.
Continuous calibration (CC) deep dive: The iterative calibration phase
From the course: Agentic AI: Build Your First Agentic AI System
Continuous calibration (CC) deep dive: The iterative calibration phase
Now, let's focus on what happens after deployment. This is Continuous Calibration, or CC, where you observe real-world behavior, understand failures, and decide when the system has earned more autonomy. Continuous Calibration is the iterative calibration phase. Now, once the system is live, even for a small group of users, you move from assumptions to evidence. This is where you learn how the agent actually behaves, not how you expected it to behave. The CC phase essentially focuses on three things. One, observing real world behavior, not just reference data. Two, identifying where the system breaks. And three, making targeted improvements. Real user behavior is non-deterministic and calibration is what bridges the gap between expectation and reality. All right, remember that we went through steps one to three in the CD phase. We start off here with step four, which is running evals on live interaction data. Depending on cost and scale, this might be a representative sample or the…
Download courses and learn on the go
Watch courses on your mobile device without an internet connection. Download courses using your iOS or Android LinkedIn Learning app.