Laurent Gil and Žilvinas Urbonas describe an AI coding workflow that selects models for tasksAI model routing sends each request to a model chosen for that request's complexity, cost, speed, privacy, or other requirements., aiming to manage cost while maintaining useful outcomes. They present internal usage and savings as their company’s own experience.
Urbonas explains a long-running workflow that breaks work into milestones, runs builds, repairs failures and scores results before stagingEvaluation is the systematic assessment of a system's behaviour or results against explicit criteria.. The speakers say human review remains part of the path to production.
They also discuss remote agent sessions and a shared team view for plans and review requests. The presentation is a vendor account of the system, not an independent evaluation.
Watch on YouTube




