Sarang Bondre
Co-founder
Eighteen years building and scaling engineering teams, most recently as a CTO-level leader at a ~$120M company — where he built the AI controls this product is based on.
Company
A company should own the way it works.
Every decision agreed, every correction made, every job that went differently than planned is an organisation working out how it operates. Today almost none of that is kept. It lives in people's heads and walks out with them, so the next team starts from nothing and each new use case is another project from scratch.
We think that should accumulate instead — held by the company, in a form it owns, while the models underneath it come and go. That is what we are building towards.
Limina is the Latin plural of limen: thresholds. A threshold is not a wall — things cross it, in both directions — and it still marks exactly where one place stops and another begins.
What we believe
Owning your AI is not about hosting the model. It is about owning the thinking around it. The model is rented and easy to swap. What should not be swappable is what your company is trying to do, what it is allowed to do, and what it has learned. Put the limits outside the model, where it cannot talk its way around them, and the rest follows.
Who we build for
We work with companies in regulated industries — finance, healthcare, legal, manufacturing and plant operations. There, the question is never whether AI can produce an answer. It is whether anyone can defend acting on it.
Team
Avinash Singh
Co-founder
AI engineer. Works on distributed systems, retrieval-augmented generation, and agent systems that hold up at scale.
Ganesh Iyer
Co-founder
Former CTO and engineering leader. Has scaled AI platforms, and the teams that run them.
Registered entity
Limina Labs Private Limited, Pune, India. We work with companies in India, the EU and the United States.