OpenAI’s India lead says companies are shifting from chatbot experiments to end-to-end AI agents that automate complex, multi-step tasks while keeping humans in the loop for sensitive decisions

Indian enterprises are moving beyond early AI experimentation and beginning to redesign workflows around artificial intelligence, OpenAI’s India lead Pragya Mishra said at an AWS event held in Mumbai on Thursday.
“When I joined in 2024, one thing was very clear to me: ChatGPT and artificial intelligence were probably going to be the most transformative technology for this generation,” Mishra said.
Mishra, who was OpenAI’s first employee in India, said the company has seen rapid enterprise adoption in the country over the last two years, with India now becoming OpenAI’s second-largest global market.
The comments came as OpenAI expanded its partnership with Amazon Web Services (AWS), bringing OpenAI models, Codex and Bedrock Managed Agents to Amazon Bedrock in limited preview. Under the partnership, enterprises can access OpenAI models including GPT-5.5 through AWS infrastructure while using existing security, governance and compliance systems.
“Enterprises want to use frontier models where their datasets, workflows, security and compliance environments already exist,” Mishra said. “With this partnership, a lot of that friction really comes down significantly.”
According to OpenAI, the integration is aimed at helping enterprises move “from experimentation to production” without rebuilding infrastructure around AI deployments. Mishra said the conversation around enterprise AI has evolved from chatbot-led experimentation to broader workflow transformation.
“The first instinct was customer support use cases, where people thought you could just plug in ChatGPT and things would start happening. Obviously not. Enterprise workflows are very complex and deeply human-dependent,” she said.
She added that companies are now beginning to focus on automating repetitive and operational tasks across departments. “If you’ve been able to make sure that a lot of repetitive and high-friction tasks are now staying on AI, that is a huge advantage both on cost and productivity,” Mishra said. “Employees can focus on creativity, domain knowledge and the passion they bring to the job.”
The partnership also includes Codex on Amazon Bedrock, allowing enterprises to use OpenAI’s coding agent inside AWS environments. OpenAI said Codex can help automate coding work, generate tests and accelerate software delivery.
Mishra also said some of the strongest AI use cases are now emerging from non-technical teams rather than engineering departments. “When we go out and do these hackathons in organisations, we see the maximum ingenuity coming from finance teams, HR teams and legal teams,” she said. “They raise their hands and say, ‘I built this finance agent which is helping me prepare audit records or financial reports much faster.’”
The companies also introduced Amazon Bedrock Managed Agents powered by OpenAI, which are designed to help enterprises deploy AI agents capable of executing workflows and multi-step tasks inside secure AWS environments. “We’re moving from question-answer interactions into the world of executing and doing things for you,” Mishra said.
She, however, noted that enterprises would still require human oversight in sensitive areas such as financial approvals and security-related decisions. “Every employee should have an AI coworker,” Mishra said, describing what she sees as the next phase of enterprise AI adoption.