Hajra Naz
Enterprise AI adoption is still shallow. Most large companies are using AI in pockets, for individual productivity, rather than wiring it into the core processes that actually run the business. That is the gap OpenAI COO Brad Lightcap pointed to in New Delhi, and it explains why the next wave of enterprise tooling is aimed at organisational workflows rather than individual chat, including the move to give AI agents their own identities inside company systems.
Earlier this month, OpenAI unveiled its new platform, OpenAI Frontier, designed for enterprises to build and manage AI agents. But despite the hype, COO Brad Lightcap told attendees at the India AI Impact Summit in New Delhi that most businesses haven’t yet fully adopted AI at scale.
“We have not yet really seen enterprise AI penetrate enterprise business processes,” Lightcap said, highlighting the gap between individual AI use and complex organizational workflows.
AI for Individuals vs. Enterprise Complexity
Lightcap explained that while AI systems are incredibly powerful for individual users, enterprises face unique challenges.
“Enterprises are highly complex organizations with many teams, tools, and goals that all need to work together. It’s not just about having AI — it’s about integrating it into systems and processes that are already complicated.”
He noted that despite buzz around AI replacing traditional software — and claims like “SaaS is dead” — the reality is different. OpenAI itself remains a big Slack user, showing that even AI-first companies rely heavily on traditional enterprise tools.
Read More: OpenAI to Prioritize Practical AI Adoption in 2026, CFO Says
Strong Demand and Revenue Growth
OpenAI CFO Sarah Friar reported in January that the company ended 2025 with over $20 billion in annualized revenue, and Lightcap emphasized that demand for AI remains extremely high.
“We almost always find ourselves managing too much demand. We’re still growing, and we’re doing our best to meet global demand,” he said.
Measuring Enterprise Success with Frontier
OpenAI is taking a new approach to measure Frontier’s impact. Instead of counting seats or licenses, the company plans to focus on real business outcomes.
“Frontier is our way to experiment iteratively, bringing AI into messy and complex business areas. If we get it right, we’ll learn a lot — both about businesses and AI systems,” Lightcap said.
Partnerships and Enterprise Expansion
Shortly after the summit, OpenAI announced partnerships with consultancies like BCG, McKinsey, Accenture, and Capgemini to help deploy AI in enterprises. At the same time, rivals like Anthropic are also rolling out Claude-based agents for finance, engineering, and design.
Lightcap also mentioned OpenClaw, a recently acquired open-source tool. While OpenAI hasn’t fully integrated it yet, the platform gives the company a glimpse into a future where agents can do nearly anything on a computer.
OpenAI in India: Growth and Voice AI
India is now OpenAI’s second-largest user base outside the U.S., with 100 million weekly ChatGPT users. Lightcap highlighted the rise of voice as a key modality, especially in regions with low bandwidth and latency.
“Voice models now feel good enough to run efficiently in environments where access to technology was previously limited,” he said, emphasizing how AI can reach previously underserved populations.
OpenAI also signed enterprise contracts in India and plans to open two offices in Mumbai and Bengaluru, likely focusing on sales and go-to-market operations. Lightcap hinted that these offices might eventually include technical teams, saying, “Never say never.”
Read More: OpenAI Opens New Delhi Office to Drive AI Expansion in India
Jobs and AI: A Changing Landscape
With AI automating tasks, there’s growing concern about job impact, particularly in India’s IT services and BPO sector. Stocks of Indian IT companies have dipped recently as investors anticipate less human involvement in areas like coding.
Lightcap emphasized a thoughtful approach:
“Over time, jobs will change. We don’t yet know how or where, but work will look different in the future. We have to show empathy for people affected by these changes.”
Common questions about enterprise AI adoption
Does slow enterprise adoption mean AI is not delivering value?
Not really. The value is showing up at the individual level, where a single person can draft, summarise or debug far faster than before. What has not happened yet is the harder part, which is redesigning a multi team process so the AI output flows into the next step without a human stitching it together by hand.
What usually stops a company from moving past AI pilots?
Three things tend to repeat: data that lives in systems the model cannot reach, unclear ownership of the process being automated, and no agreed way to measure whether the pilot worked. Pilots that fix a narrow, well instrumented workflow are the ones that tend to survive review.
