TLDR: A panel at TechSparks 2025 discussed the transformation of AI from simple assistants to autonomous collaborators in the workplace, automating decision layers across various sectors. Leaders from CoRover.ai, Observe.AI, and DevRev highlighted the shift towards AI agents that understand intent, take action, and continuously learn, emphasizing human-AI collaboration over replacement. The discussion also covered the challenges of implementing AI in enterprise environments, focusing on data privacy, security, and the critical need for accuracy to build user trust. Companies are developing specialized AI models for proprietary data, achieving high accuracy rates in critical applications without human intervention. The future of work is seen as a partnership where AI enhances human capabilities and adaptability becomes the new measure of productivity.
At TechSparks 2025, a distinguished panel featuring Ankush Sabharwal, Founder & CEO of CoRover.ai, Jithendra Vepa, Co-founder & CTO of Observe.AI, and Manoj Agarwal, Co-founder & President of DevRev, delved into the profound impact of agentic AI on work and collaboration. The central theme revolved around AI’s evolution from mere task executors to autonomous collaborators, quietly automating decision-making processes traditionally handled by humans across various enterprise functions, from finance to HR.
Manoj Agarwal of DevRev highlighted the distinction between general-purpose AI like ChatGPT, trained on public data, and the requirements of enterprise AI. He emphasized the challenge of creating a ‘ChatGPT-like experience’ within the secure, private, and compliance-governed environment of an enterprise. DevRev’s mission, he explained, is to build systems that enable teams not just to search but to ‘reason with enterprise data, take action on it, resolve issues, and collaborate more intelligently,’ ensuring AI-powered collaboration is secure, context-aware, and enterprise-grade.
Jithendra Vepa, from Observe.AI, shared his company’s journey in transforming contact centers using AI. Initially focused on objective analysis of 100% of customer conversations, Observe.AI expanded to real-time assistance. Vepa noted that with advancements in Large Language Models (LLMs) and Retrieval-Augmented Generation (RAG), AI agents can now provide instant, accurate answers, moving beyond reactive chatbots to proactive, continuously learning collaborators that understand intent and take action.
Ankush Sabharwal of CoRover.ai articulated his company’s evolution from promoting chatbots to offering a full-stack enterprise AI platform. He stressed that the focus should be on ‘solving real business problems’ rather than buzzwords. CoRover empowers enterprises to build small language models on their proprietary data, offering prebuilt solutions for sales, support, and workflow automation through its Generative AI Care platform. The company recently set a world record by creating 7,000 AI agents on its platform in a single day, surpassing Google’s previous record of 2,000. Sabharwal underscored that the true excitement lies in helping enterprises achieve their purpose, even by a small margin.
The panel also addressed the critical aspect of measuring productivity in an ‘agentic AI world.’ Agarwal pointed out that unlike the deterministic nature of traditional automation, where repeatability was key, the new paradigm of agentic AI values ‘adaptability.’ He stated, ‘We’re moving from machines that follow rules to systems that reason.’ This transition involves AI initially working in ‘assist mode,’ gradually shifting to ‘auto mode’ as trust develops, leading to human-AI collaboration where both define priorities and route tasks, fostering ‘true productivity.’
Vepa echoed this sentiment, viewing AI’s evolution as cooperation rather than replacement, with shared ownership in customer interactions. He highlighted Observe.AI’s rigorous testing of AI agents through synthetic conversations to ensure predictability and identify biases in this ‘probabilistic paradigm.’
Also Read:
- Generative AI’s Transformative Impact on India’s Business Workflows
- Tata Consultancy Services Details How Agentic and Generative AI Drive Enhanced Decision-Making in Consumer Packaged Goods
Addressing concerns about AI’s limitations and the risk of hallucination, Sabharwal firmly stated that CoRover.ai operates with a ‘zero tolerance’ policy for hallucination in B2B applications. He emphasized that for critical sectors like banking or train bookings, ‘there’s no room for error,’ and CoRover agents ‘prefer not to answer rather than hallucinate,’ prioritizing trust and accuracy. He also revealed that solutions built for major Indian entities like RBI, NPCI, IRCTC, SEBI, and LIC operate without human handovers, maintaining over 95% accuracy. Sabharwal concluded with a poignant example of AI’s human potential, citing a student who built an agent to preserve his late mother’s teachings, illustrating AI’s capacity for deeply human applications when built with the right intent.
