Krutrim unveils India's first agentic AI assistant - Kruti
Krutrim, the artificial intelligence company founded by Bhavish Aggarwal, has announced the launch of its new agentic AI assistant, Kruti. The Bengaluru-based startup confirmed on Monday that the product will be released on 12 June, positioning it as "India's first agentic AI assistant."
In a post on X, the company said Kruti "listens, adapts and acts proactively, purposefully and in your language". The message added: "This is a leap beyond chatbots. More updates on 12th June. Stay tuned!"
The announcement comes months after Aggarwal first teased the development of Kruti. In April, he praised the team behind the assistant, noting their efforts to build "a product as good as the global apps, but better for the Indian users."
Kruti will be available initially to consumers, with a version for developers to follow, allowing them to create and deploy new agents on the platform. Navendu Agarwal, head of business at Krutrim, and Chandra Khatri, the company's founding head of AI, said Kruti's capabilities range from cab booking and food ordering to bill payments and image creation.
"It delivers task execution – from cab booking, food ordering, bill payments to image creation," Agarwal said. He added that the AI also supports advanced features such as in-depth research, with all functions currently available free of cost to users.
Agarwal explained that Krutrim intends to begin by addressing complex use cases within Ola's ecosystem before expanding to other services. He highlighted that Kruti can understand both voice and text commands, recall past interactions, and work in over 11 Indian languages. "Whether it's 'book me a cab to the airport' or 'order my usual lunch,' Kruti understands intent and gets the job done," he said.
According to the company, Kruti's design focuses on the Indian market, with features tuned for local voices, dialects and bandwidth constraints. Powered by Krutrim's proprietary V2 large language model alongside open-source models, it aims to deliver scalable and cost-efficient performance.
Aggarwal described the release as "the first real step towards the future of AI where technology doesn't just talk back, but actually helps you get things done." The assistant is intended to learn user preferences over time, adapt to usage patterns, and integrate with a variety of apps and services to deliver context-aware assistance.
Kruti includes three modes similar to those seen in other AI chatbots: 'Instant' for quick responses, 'Insightful' for reasoning tasks, and 'In-depth' for research. The system is optimised for smartphones and will initially connect to Ola ecosystem services, with plans to expand integration to partners such as Blinkit, Swiggy and Uber, alongside voice command support.
The AI assistant also offers a software development kit for third-party developers, enabling platforms to integrate its orchestration, memory management, and task execution features with minimal coding.
The launch comes as Krutrim faces challenges with some of its other offerings. Earlier this month, reports suggested its large language models and cloud products had struggled to gain traction, with some developers describing them as lacking in technical maturity compared to larger providers.
Krutrim became India's first AI startup to reach unicorn status in 2024 after securing $50 million in funding at a $1 billion valuation. In February this year, Aggarwal announced plans to invest Rs 2,000 crore in Krutrim, with a further commitment of Rs 10,000 crore by next year.
Industry observers have noted growing interest in agentic AI, with Deloitte recently launching its Asia Pacific Agentic AI Centre of Excellence in India, Malaysia and Singapore. Deloitte said such systems enable "autonomous agents to independently manage complex workflows, learn continuously and adapt over time."
Kruti's debut on 12 June will mark a significant step in Krutrim's expansion into the competitive AI assistant space, with the company emphasising its focus on local relevance, language diversity and task-oriented performance.