New Delhi, 30 September, 2026: Pienomial today launched AT0M, an AI model built for the small, repeated decisions that run through every business: which team should handle a request, whether an alert needs attention today, how serious a reported problem is. Because it never generates free text, every response stays within the options the developer defined. Teams can automate routine cases and send uncertain or high-stakes ones to a person for review.
Unlike hosted AI services, AT0M is a model organisations own outright. It trains and runs on their existing hardware, their data never leaves their environment, and there are no per-seat or per-token charges.
“For the past decade, businesses have been told the only way to use AI is to rent it: send your data out, pay by the token, and trust someone else’s model,” said Sanat Mohanty, CEO of Pienomial. “Most business decisions are small, well-defined, and repeated thousands of times a day. They don’t need a giant hosted model. They need something reliable that you own and control. That’s what AT0M is.”
AT0M runs as a single executable with no network dependencies, on standard processors, including Intel® Xeon® and hardware such as Apple Metal and NVIDIA CUDA. That makes it suited to organisations where data cannot leave the premises, including banks, government agencies, and healthcare and life sciences companies.
AT0M is a System One model, part of a fast-emerging category of AI, popularized recently by TypeSafe’s Jev, that makes structured decisions for software rather than generating text.
In company-run testing on the public typed-decisions benchmark, AT0M matched the reference answers on 78.9% of 2,000 decisions, compared with 74% reported for TypeSafe’s Jev. AT0M produces deterministic, reproducible responses. In a hosted test on an NVIDIA T4 GPU, AT0M answered in under 16ms with comparable calibration error, making it ready for use in enterprise settings.