Srinivasateja Songa, Staff Software Engineer, The Home Depot, Atlanta, USA

Title of the Talk :
Effective Context Management for AI Agents: Leveraging Chat History to Reduce Hallucinations and Optimize Token Usage

Abstract of Talk:
Large Language Model (LLM)–powered AI agents are increasingly used in conversational systems, enterprise automation, and intelligent applications. However, maintaining accurate context across long conversations remains a major challenge. As interactions grow, naïvely including entire chat histories in prompts leads to excessive token usage, reduced efficiency, and a higher risk of hallucinations.
This keynote explores strategies for effective context management in AI agents by intelligently leveraging chat history. The talk discusses architectural approaches that combine conversational memory, selective context retrieval, and token-aware prompt construction to maintain coherent multi-turn interactions.
By prioritizing relevant historical information, AI agents can significantly improve response accuracy while reducing token consumption. The session will also highlight emerging trends in conversational memory and agentic AI architectures that support the development of more reliable and scalable AI-driven systems.

Bio : Srinivasateja Songa is a Staff Software Engineer and enterprise technology professional with over 15 years of experience in cloud architecture, enterprise software development, and AI-driven systems. He currently works at The Home Depot, where he contributes to the design and development of scalable cloud platforms and intelligent enterprise applications.
His expertise spans enterprise CRM architecture, distributed systems, cloud integrations, and AI-enabled automation, with a strong focus on building scalable and reliable technology platforms. Srinivas has extensive experience working across industries including retail, finance, healthcare, and life sciences.
In addition to his industry work, he actively contributes to the research community as a peer reviewer for IEEE conferences, supporting advancements in artificial intelligence and software engineering. His research interests include Generative AI, AI agent architectures, conversational memory systems, and enterprise AI applications.

.