Self-learning agents

Give an agent an objective and a metric. Each run it makes one focused change, measures the result and writes what worked into a persistent playbook it reuses next time.

• One change per run

  Each run makes a single focused change rather than a random rewrite.

• Tracked metric

  Set the metric and baseline you care about; every run measures the improvement.

• Persistent playbook

  Lessons from past runs are written into a playbook the next run starts from.

• Full run history

  Enable or pause any agent and review every run, its change and credit cost.