Over the last few months, Indian fintech is getting the AI glow-up.

Different firms are at different stages of implementation. But mind you, very few of the AI use cases particularly blow your mind.

As the CTO at a fintech company says, “I think in sectors like healthcare, we have seen far more exciting use cases than in financial services, which does not go beyond recommendations and portfolio analysis.”

Now, both the stock market and banking regulators are doing their own research into how to regulate the use of AI. Sebi in June had a consultation paper inviting comments on responsible use of AI. The RBI, too, in August conducted a survey among 600 financial entities on their use of AI and laid out an exhaustive set of 26 early recommendations on AI implementation.

All things considered, the choice that different fintechs make has a lot to do with their risk appetite—and right now, they are spread across five stages.

These are the low risk takers. These companies have way more to lose than to gain by integrating AI into their consumer-facing systems. A clear instance of this is when India’s largest UPI-payments app Phonepe partnered with OpenAI this month to bring ChatGPT to Phonepe users. Essentially, OpenAI hopes it can get Phonepe’s 600 million users to use ChatGPT.

If you read the announcement around it, it sounds like a big deal. “This partnership will demonstrate the immense value of consumer AI across India, helping millions of users enhance their daily lives.”

But essentially, it’s a way for Phonepe to make some ad revenue off of OpenAI. I’d reckon it’ll take a lot for Phonepe to wade into the stickier territory of integrating AI into its customer-facing interface, or experimenting with any swashbuckling new feature that will warn you against making a payment to a friend who never pays you back.

This is where most companies are at now, and it’s not unique to fintech. Companies are using foundational models or AI tools to tune different parts of their operations—from building better cybersecurity features to getting their engineers to accelerate development cycles. About 47% of Indian companies have reported moving their generative AI pilots into live, functioning use cases. And about 23% are currently experimenting with these products, according to a recent EY report.