Free Lesson
Your AI agent works. Your bill doesn't.
45 min
Aug 3, 2026 11:00 AM
Virtual (Zoom)
In this video
What you'll learn
Where the money actually goes
Measure cost per successful task rather than cost / API call. Identify three items that quietly dominate most agent bill
Model routing that protects quality
Route requests to a lower-cost model first and escalate based on confidence. Understand the decision logic & guardrails
Caching, batching and context discipline
Apply the practical techniques that reduce spend before changing models. Learn how to determine which technique works
Cost-aware evaluation
Track quality and cost in the same evaluation harness so that cheaper does not quietly mean worse.
Why this topic matters
Most agent projects don't die from bad models:
They die when someone in finance asks what the system costs per successful outcome and the team has no defensible answer. Inference spend is an architectural decision - not simply a bill that arrives.
In this session, participants should be able to model an agent's unit economics, select a routing strategy using evidence and defend the architecture
You'll learn from
Dr Ankur Narang
Dr. Ankur Narang brings 30+ yrs exp in AI, Tech across MNCs & many verticals
Kush Khurana
AI & ML leader; Venture Partner, DeepCoreX; Ashoka faculty
IBM; Apparel Group; Yatra; Hike
