Enterprise AI in Production

Session Description

Most enterprise AI agents that fail in production don’t fail because of model quality - they fail because they have no ground beneath them. Without structured semantic context, agents hallucinate over ambiguous schemas, misfire on intent, and produce answers that are technically fluent but operationally wrong. This session walks through the full grounding stack - from semantic metadata and knowledge graph design to hybrid Text-to-SQL and RAG pipelines - drawing on a real deployment built for a Fortune 500 client’s enterprise data platform. We’ll examine how a multi-agent system was constructed on top of complex relational infrastructure, how a knowledge graph was introduced to give agents the contextual backbone they needed to reason correctly across domains, and how a query validation gate was added after observing the failure modes that emerge when agents are trusted without checks.

The session goes beyond architecture slides. We cover the operational decisions that only become visible under real workloads: why a custom orchestration approach was chosen over a managed platform, how intent classification was layered in before query generation, where the RAG and structured-query paths conflict and how those conflicts were resolved at runtime, and what the tracing and observability layer revealed about agent behaviour that offline testing never surfaced. Attendees will leave with a concrete implementation model - not a framework pitch, but a set of grounded architectural patterns built around knowledge graphs, semantic injection, and evaluation signals that transfer across domains and industries.

Key Takeaways

1. A repeatable grounding architecture - covering knowledge graph design, semantic metadata injection, intent classification, query validation gates, and RAG-structured-query conflict resolution - that you can adapt for any enterprise AI agent operating over complex or semi-structured data.

2. A practical failure taxonomy drawn from production traces: the specific points where ungrounded agents break, how observability tooling surfaces those failures, and which architectural interventions - including knowledge graph grounding - actually reduced hallucination and improved reliability at scale.

Target Audience

This session is most valuable for System Architects designing or operating multi-agent systems in enterprise environments, platform. Backend engineers integrating LLM-based capabilities with existing data infrastructure, and engineering leaders making build-vs-buy and infrastructure decisions around GenAI platforms.

Anyone asking “how do we make this reliable in production?” will find direct, transferable answers here.

Speaker Bio

Nabendu Karmakar is a technology leader and product builder with over 14 years of experience at the frontier of Agentic AI, Generative AI, Data and Fullstack Engineering. His career has been shaped by a single question most teams skip: will this system still work a few months later, inside a messy enterprise, with real data and real constraints?

He currently serves as Principal Architect at Fractal Analytics, where he leads a team of engineers building large-scale data and AI platforms for Fortune 500 clients across manufacturing, insurance, CPG, and pharma. His work spans multi-agent systems, knowledge graph-driven architectures, Text-to-SQL engines, and RAG pipelines - always with a consistent focus on systems that are operationally sound, architecturally honest, and built to deliver measurable impact in the real world.

Before Fractal, he was part of Jio Haptik - one of the world’s largest conversational AI platforms - and began his career at L&T Infotech, one of India’s most respected technology firms.