When you launch your improved $BORT agent, the metrics panel works like a feedback system showing how well your agent is performing and learning.

Total Interactions increments whenever end users query the agent. A surge after adding knowledge means the agent is being used more, possibly because it now answers new categories of questions.

Learning Events track internal updates successful ingestions, fine tuning steps, or reinforcement learning batches triggered by fresh data. Each approved Knowledge Source typically triggers at least one event, so you should see a jump soon after upload.

Confidence Score (0‑100%) estimates response reliability. When newly added knowledge fills previous gaps, you’ll notice gradual gains; if conflicting or low quality data is added, the score can dip, signaling a need to adjust priorities or clean the source.

Learning Velocity visualizes the slope of improvement across recent interactions. A spike after knowledge ingestion implies the agent is assimilating the material quickly; a flat line suggests minimal behavioral change.

BNB Balance reflects the operational budget powering on-chain actions (storing checkpoints, agent to agent calls). Uploading knowledge alone doesn’t move this number, but heavier usage driven by better answers will consume funds, so you may need to top it up via Fund Agent.

Monitoring these indicators after each knowledge upload lets you verify that the agent is learning what you intended, catch regressions early, and fine tune priorities or content quality for sustained growth.