Day one of MLDS 2025, India’s biggest GenAI summit for developers, hosted by AIM Media House, was a day filled with energy, excitement, and forward-thinking discussions. The day witnessed the presence of close to 1000 attendees.
One of the day’s interesting talks came from Rahul Bhattacharya, AI leader at GDS consulting, EY, who delved into the world of the agentic workforce, the next frontier in AI-driven workspaces. Bhattacharya highlighted how assessing the risks of integrating AI agents into the workforce is just as crucial as measuring the benefits.
He broke down what makes a system truly “agentic.” To be considered an agent, Bhattacharya explained, a system must have the ability to interact with its environment, make decisions based on its observations, and learn from its actions to continuously improve.
He elaborated, “A key ability is making decisions, where the agent chooses the best action based on set rules, goals, or rewards. Over time, it should learn from past experiences and feedback to improve its performance.” This blend of adaptability and decision-making could redefine how we think about work in the future.
In the digital content world, where video has taken center stage, Arvind Sasikumar, co-founder and CTO at Quinn, shared his insights on the critical importance of video compression.
Sasikumar emphasised that optimising video transcoding isn’t just about reducing file sizes, but it’s about enhancing the user experience by eliminating buffering, which can be a dealbreaker for viewers. He explained how even a 5% reduction in file size can make all the difference for users with lower internet speeds.
“Consider this: If 100 users have a 1.9 Mbps connection but the video they’re watching requires a 2 Mbps bitrate, every single one of them will face buffering. But by optimising compression, we can ensure that all 100 users enjoy smooth playback,” Sasikumar said.
As organisations race to integrate AI into their business strategies, there are hurdles to overcome. Chirag Jain, vice president of AI Practice at Genpact, took the stage to shed light on why businesses must first define their long-term goals before jumping into AI implementation.
He compared it to the vision of a sports team– just as India’s cricket team aspires to be the best in the world, businesses also require a clear, long-term goal to guide their AI strategy.
Jain underscored that a strategy grounded in core values, ethical considerations, and regulatory compliance is key to ensuring AI’s success within an organisation. “Nobody likes to see a team playing unfairly,” he said, reinforcing the need for transparency and ethical decision-making in AI adoption.
The day also featured Siddhant Goswami, co-founder at 100xEngineers and Tech Influencer, who spoke on the evolving role of AI agents. While AI agents are becoming increasingly autonomous, Goswami reminded the audience that human intervention is still essential.
“We can’t completely rely autonomously on them. We need human feedback. These agents are self-directed, capable of planning and executing multiple steps based on feedback and updated goals,” Goswami shared.
The excitement continued with a workshop by Gopala Dhar, an AI engineer at Google Cloud, and Lavi Nigam, a developer relations engineer at Google Cloud, on building real-time applications with the Gemini Multimodal Live API. This hands-on workshop served as a holistic guide to using Gemini’s multimodal capabilities to create sophisticated applications that can see, hear, and interact naturally.
Also, the day witnessed several paper presentations, including Life Stage Customer Segmentation by Fine-tuning Large Language Models by Nikita Katyal, Head of Analytics and AI at Central Retail Corporation, and Vivek Vishwas Vichare; A Context-Aware Multi-Agentic Multi-Modal LLM Architecture for Digital Marketing by Vivek Vishwas Vichare, head of data sciences and analytics at Pixis, and more.
These papers showcased the innovative ways AI and generative models are reshaping industries like retail, digital marketing, and healthcare, demonstrating the immense potential of AI to solve complex problems across various domains.
Day one of MLDS 2025 left attendees with plenty to think about.
