Bharani Subramaniam

Bharani Subramaniam

CTO, India and the Middle East Thoughtwork

Bharani, with 24 years of experience has been building applications since the late nineties, when he started his career as a VC++ win32 programmer. He is a connoisseur of various technologies but his passion lies in simplifying system designs. He also indulges in building strategies around data engineering and making the most of real time analytics for Thoughtworks’ clients.

Bharani in his current role shapes new pursuits and builds tech strategy with clients. He collaborates with the Thoughtworks' Technology Advisory Board that puts together thoughts on emerging technology trends in the industry, resulting in the much sought after Technology Radar.

Earlier leadership roles that Bharani has held, include being the Head of Technology for India, and Office Technology Principal for Thoughtworks Chennai.

While, Bharani’s geeky passion lies in learning and building low latency, high throughput applications, he has also capitalized on opportunities to design and build distributed systems, leveraging microservices. And what he loves most about coming to work is being surrounded by people who are smart and contribute to Thoughtworks’ open culture, which he believes, provides a great platform for people to question, challenge and learn.

The technology leader’s hobbies outside technology, include playing a mean game of badminton and watching movies that are from the Marvel Cinematic Universe.

Keynote Title

Emerging patterns in Agentic AI


Session Overview

This talk explores the rapidly evolving landscape of Agentic AI systems, beginning with fundamental definitions that distinguish agents from other AI paradigms. We'll examine the diverse continuum of real-world AI agents, highlighting how they vary in autonomy, capabilities, and application domains.

We will analyze critical enabling technologies, including the Model Context Protocol that facilitates sophisticated agent-environment interactions. We'll delve into action spaces—with special attention to code generation as a particularly powerful action domain—and explore how tool integration is creating unprecedented ecosystem opportunities.

Memory plays a crucial role in context management and agent learning. We will explore different memory architectures, from short-term recall mechanisms to long-term persistence strategies, showcasing how they enable truly agentic behavior.

The session concludes with a forward-looking perspective on the future of AI agents, including emerging model architectures designed specifically for agent-based applications. By mapping these emerging patterns, attendees will gain both a conceptual framework for understanding the current state of Agentic AI and insights into future advancements.

Key Takeaways

  • • Understanding Agentic AI – What sets AI agents apart from traditional LLM-based automation.
  • • Model Context Protocol – How context-aware AI agents interact with their environment.
  • • Memory Architectures in AI Agents – From context management to long-term learning.
  • • Action Spaces & Code Generation – The role of tool integration in expanding AI capabilities.
  • • Future Trends in Agentic AI – The shift towards purpose-built AI architectures for agent-based systems.
About Us
EVENTS
Resources
Policies