The Fifth Elephant Pune edition conference
summary
Claims a deployed system that turns LLMs into a decision engine for smart buildings, not a chatbot: it uses stateful orchestration, a persistent context graph and LangGraph-style tool routing to read telemetry, query APIs and write recommendations. Built components: event ingestion, context store, agent runtime, tool schemas and guarded executors; it handles thousands of sensors and millions of events. Reported wins: operators shift from data-collection to action-evaluation, decision stability rises and recommendations get consistent approvals; still wrestling with memory sync, causal correlation and latency vs. depth — evaluation now weighs routing precision and override rates, not just model accuracy.
AI-written summary · about
chapters (18)
- 0:00Conference opening
- 0:28Fifth Elephant Pune intro
- 1:09Evolution of Fifth Elephant
- 2:13Pune conference planning
- 4:02Conference schedule
- 5:55Morning session begins
- 16:36Agentic intelligence for buildings
- 54:25Causal inference in manufacturing
- 1:37:02AI-driven SDLC at Travelopia
- 2:27:10Agentic QA at Nutanix
- 3:00:16AI grants for Indian startups
- 3:06:19AI systems for software dev
- 4:31:54Autonomy 2.0 for driving
- 5:00:16AI test automation at IDfy
- 5:38:56AI for e-commerce catalogs
- 6:14:09Flash talks: AI projects
- 7:22:51Agentic AI for flight scheduling
- 8:12:21Conference closing Q&A
related talks
31mAI Agent Orchestration for Enterprise Arch with AWS Bedrock, Flyte & LangGraph
41mBuilding LLM Agents – Challenges and Learnings
16mBeyond LLMs - The Next Frontier of Machine Reasoning
33mOpening Keynote: LLMs, between fantasy and reality - the poin...
35mBuilding LLM-based Agents
48mMastering the AI Shift: Adaptive Strategies for Software Evolution
48mUnlocking LLM Potential: From Exploration to Integration
14mEdge of the Present: Pursuing AI’s Frontier in the Enterprise