She said that companies are increasingly deploying AI in real-world applications, prioritizing trustworthiness and trustworthiness rather than just model functionality.
“As you build, think about solving real problems in India, especially the last-mile challenge, while also contributing to broader global outcomes. The opportunities that are set are immense,” Ghose told developers while speaking at a hackathon event for vibecoding AI startup Emergent.
The startup is backed by Khosla Ventures, SoftBank, and YC and is valued at $300 million after a $70 million round in January.
Anthropic, which is currently attracting investor interest with a valuation of $800 billion, counts India as a key market and said adoption cycles are accelerating and businesses are no longer questioning the relevance of AI. “They feel the bus is coming and they decide when to board,” the MD said during a fireside chat, adding that organizations no longer need to justify the case for AI because that question has already been answered.
Ghose said India’s strength lies in its diversity, scale and ability to solve last-mile challenges. “If we can solve these constraints, we can build solutions that are relevant not only for India but for many other parts of the world.”
She urged builders to focus on harder problems rather than barriers to entry. “Focus on solving the biggest and most complex problems, the ones that haven’t been solved yet.” India’s MD told developers that barriers to entry have already come down significantly.
Selective approach to mythology
For initiatives like Project Glasswing, which Anthropic announced in early April as part of its latest model, Claude Mythos Preview, the company has chosen not to release the model to the public.
The company has formed a limited consortium of approximately 40 to 50 organizations that have been granted early access to the model to test and ensure the security of the system before similar functionality becomes widely available.
“The aim is to work with a broad set of partners to test and validate functionality in a controlled manner,” Ghose said, adding that the aim is to ensure that the performance and functionality of the model is robust, especially in areas such as cybersecurity.
“We will continue to expand our scope in future stages while learning from our initial set of partners.”
She added that Anthropic has already released support resources for developers and plans to expand the scope in future stages. “We are already publishing capabilities around AI for cybersecurity through blogs and whitepapers that developers can still explore and implement.”
The company works with government priority sectors such as education, skills training, agriculture and healthcare. Ghose said Anthropic is working with developers and the broader ecosystem to build “from India and for India.”
Regarding the artificial general intelligence (AGI) timeline, he said the timeline is becoming compressed given how quickly models are improving. At the same time, as capabilities improve, it is equally important to ensure responsible implementation.
Also read: From AI to everything: How Anthropic is defending the tech legacy
