About
AI that can be checked against the evidence
Scaling Truth Research Foundation is a not-for-profit AI research lab based in Gurgaon, India. We build models whose outputs can be traced to evidence and tested by others, publish how we evaluate them, and bring the work into use through products built by OneThought AI.
- Legal form
- Not-for-profit research organisation
- Based in
- Gurgaon, Haryana, India
- Research areas
- Life sciences, realtime video, media intelligence, software from intent
- Contact
- connect@scalingtruth.org
Mission
AI now produces fluent answers in every field, and fluency is no longer a sign that an answer is right. Our mission is to build AI whose outputs can be checked: traced to their sources, honest about uncertainty, and evaluated in ways others can repeat. We start where a confident wrong answer is costly, in health, in the news and in the software people rely on.
Why a not-for-profit
Some of the most useful work in AI is the least profitable: evaluations that show where a model fails, datasets built for public research, models for questions no product company would prioritise. A not-for-profit can do that work and publish it, and hold the mission steady when commercial pressure points elsewhere.
Why India
India holds deep knowledge in the life sciences, including Ayurveda, in texts, laboratories and practice that general-purpose models do not encode. It also has a research base able to build AI capability rather than import it. We work from India, for Indian researchers, institutions and patients, and publish for the world.
How the work reaches people
Research produces models, methods and evaluations. Where the work is ready, OneThought AI builds products on it: Media Studio, Media Monitor and Viber. Revenue and what we learn from real use return to the research. Structure and governance explains the relationship and its rules.
What we will not do
We do not make diagnostic claims, generate the likeness of real people without consent, build tools to watch private individuals, or train on data we have no right to use. The full list is in our safeguards.