DeepTech Entrepreneur, Technology Leader
AI & ML Leader, Venture Partner - DCX

.jpeg&w=768&q=75)
This workshop matters because the hardest part of AI agents is no longer getting them to work once — it is making them reliable, safe, traceable, and controllable when they interact with real systems. A demo may succeed on the happy path, but production agents must handle missing evidence, tool failures, retries, policy violations, approval requirements, and unsafe model recommendations.
Build. Break. Prove. gives participants a hands-on way to experience those realities. They will build a production-minded agent, deliberately expose it to failure, inspect the evidence, and repair a real weakness. The workshop teaches a critical principle: the model can propose, but deterministic controls must decide what is valid, safe, and releasable.
Participants leave with a working prototype, execution trace, diagnostic scorecard, and a clear view of what still separates their agent from production readiness. It is designed to move developers beyond prompt engineering and agent demos toward the engineering discipline required for dependable enterprise AI systems.
You won’t just learn why AI agents fail in production - you’ll work on patterns that make them usable, reliable, and cost-aware.
Learn to turn AI agent demos into reliable, evaluated, cost-aware production systems, and gain the judgment to lead real deployments.
Build an agent that handles retrieval, reasoning, tool use and structured outputs.
Test the agent against edge cases and refine it
Finish with a deployable, demo-ready agent for a real business use case.
Add production-grade RAG with dense, sparse or hybrid retrieval patterns.
Measure retrieval quality using precision, recall and source-grounding checks.
Detect stale, missing or weak context before it causes unreliable responses.
Route tasks between smaller and frontier models based on complexity and value.
Apply caching, batching and context optimization to reduce inference costs.
Produce a cost and architecture brief explaining model choices and trade-offs.
Handle model, tool and retrieval failures with explicit fallback paths.
Use human-in-the-loop escalation when confidence or system state requires it.
Manage workflow state so failures do not force unsafe or unnecessary restarts.
What changes when an agent moves from demo to production
Reasoning, tools, structured outputs and reusable agent patterns
RAG patterns, source grounding, retrieval quality and failure modes
Choosing frontier vs smaller models, caching, batching and cost optimization

Dr. Ankur Narang brings 30+ yrs exp in AI, Tech across MNCs & many verticals

.jpeg&w=384&q=75)
AI & ML leader; Venture Partner, DeepCoreX; Ashoka faculty


BTech & MTech IIT Mumbai. AI Lead with solid Agentic Experience

AI/ML engineers & developers who can build agent prototypes and want to make them reliable, measurable and production-ready.
Product, engineering & AI leaders responsible for deploying agents safely, cost-effectively and at scale in real workflows.
Founders, technical and non-technical professionals building or adopting Agentic AI who need practical patterns, risks and value clarity.
Live sessions
Learn directly from your instructors in a real-time, interactive format.
Lifetime access
Go back to course content and recordings whenever you need to.
Community of peers
Stay accountable and share insights with like-minded professionals.
Certificate of completion
Share your new skills with your employer or on LinkedIn.
Maven Guarantee
Your purchase is backed by the Maven Guarantee.
Maven for Teams
Reimbursement
Get your company to pay
Everything L&D needs: email template, receipts, and certificate of completion.
Get reimbursedTeam discount
Learn with your teammates
Save 20%+ when 2 or more teammates enroll in the same cohort.
Save 20%+ with a teamPrivate cohort
Run a cohort for your org
A dedicated cohort with a custom schedule and curriculum, tailored to your team.
Book a private cohort$200
USD
4 days left to enroll
11am–1pm EDT