Agents are entering the systems that decide credit, claims and investments for millions of Indians. Rotavision exists so that every one of those decisions has an owner, a record, and evidence a regulator can trust, built for India, from India.
Why we exist
The tools for building AI agents are racing ahead of the tools for governing them, and the governance that does exist is written for other regulators, in other languages, for other categories of harm. India's rules are its own: RBI model risk management, the DPDP Act, SEBI's responsibility provisions, IRDAI on the horizon; fairness framed around gender and caste, not the categories a US-built tool ships with. Sakshi is the accountable record beneath the agents you already run, encoded for those instruments, deployed inside your environment, honest about what it can and can't yet prove.
What we believe
Not a foreign product with a rupee sign bolted on. The regulatory mappings, the fairness categories, and the residency model are built for India as the primary case.
We don't ask a regulator to take our word for anything. Every claim is a hash-anchored record they can recompute themselves: proof, not a dashboard you're asked to believe.
Sakshi is single-tenant and runs in your environment. We never see your evidence and couldn't hand it over if compelled. PII is tokenized before it is ever stored.
A clause reads READY only when a real check backs it. An honest gap you can close beats a green box you can't defend. That discipline is the whole point of the product.
Engaged, not observing
We don't wait for directions to land and then scramble. Rotavision filed a formal comment on the RBI's draft model-risk-management directions, and tracks each instrument as it moves, so the day the final rules publish, our customers are already scored against them.
Building the governance layer for Indian finance means being a serious participant in how that governance is defined.
Backed by open research
Sakshi's governance models, PII detection, India-calibrated fairness, and groundedness judging, come from open research at Rota Labs, published Apache-2.0 on Hugging Face for anyone to inspect. Calibrated for Indian data and code-mixed language, not adapted from someone else's.
The entity
Recognized by Startup India (DPIIT), Bhaskar ID OI-0226-9390LW.
Work with us
Teams putting agents into real workflows, shape the product and get ahead of what your regulators and auditors will ask for.
hello@rotavision.comA private, synthetic-data sandbox in minutes, register an agent, witness a decision, generate an evidence pack.
Request a sandboxQuestions, partnerships, or media, we read everything that comes in.
hello@rotavision.comExplore the live console with synthetic data, or talk to us about a design partnership.
