We’ve all been there, waking up the morning after a lavish wedding feast and realising you never made it to the gulab jamun or that live pasta counter. Indian wedding buffets are legendary for their scale, but also notorious for overwhelming guests who end up full before tasting half the spread. Now, a Bengaluru-based software professional claims to have found a tech solution to this very Indian problem.

Meet BuffetGPT, an experimental AI tool created by Pankaj, a techie who says years of food regret finally pushed him to act. “Indian wedding buffet is a scam,” he joked online. “I always leave regretting something.”

So what did he do? He built an AI agent that scans the entire buffet and creates a personalised eating strategy, telling guests what to eat first, what to skip, and how much to take, all so they can enjoy more dishes without feeling stuffed.

How does BuffetGPT work?

Using computer vision, BuffetGPT identifies every dish on display. It then applies what Pankaj calls “actual stomach volume physics” to calculate the best combination and sequence of items, helping users maximise variety while avoiding the post-buffet food coma.

In simple terms, it’s like having a smart food coach standing next to the chaat counter, whispering, Go easy on the biryani — the rasmalai is coming.

The tool is still in its early stages and has only been tested in alpha mode at one of his friends’ weddings. But Pankaj says the results were “decent,” with users managing to sample more dishes than usual without overindulging.

Signing off his post with humour, he wrote, “Honestly, this is what my computer science degree was for.”