An applied AI research lab

Research that ends up in someone's hands.

We build the capability first: reasoning, grounded retrieval and adaptive systems. Then we take it into one domain at a time and ship a product real people depend on.

Founded2026, Mumbai
TeamIIT Bombay
First domainEducation
StatusFirst release soon

Capability tracks

Domain independent
01Active

Reasoning & verification

Multi-step problem solving where each step can be checked, not just plausibly worded.

02Active

Grounded retrieval

Answers tied to real source material, with every claim traceable back to it.

03Active

Adaptive user models

Inferring what someone actually misunderstands from the way they get things wrong.

04Exploratory

Efficient inference

Making capable models cheap enough to run for millions of requests a day.

Where we apply it

The same core work carries across fields. We enter a domain only when we can build something measurably better than what exists there.

First domain

Education

Personalised learning support, starting with competitive exam aspirants.

Exploring

Scientific tooling

Assistants that read the literature and hold a line of reasoning together.

Exploring

Knowledge work

Systems that answer from an organisation's own material, with citations.

Next

Yours, possibly

We partner on hard problems where verification actually matters.

First release / EducationComing soon

A tutor that starts where you're stuck.

A personalised doubt-solving platform, starting with GATE CSE aspirants. Ask in your own words, get a worked explanation at your level, and see the concept behind the mistake, not just the answer.

GATE CSE is where it opens; more exams follow. The first surface of the education work, not the whole of it.

Step-by-step solutions, checked for correctness
Learns which concepts you keep slipping on
Practice generated around your weak spots
ai tutor

A worked derivation, built one step at a time — not just the final answer.

Who we are

A small team of researchers and engineers from IIT Bombay, working at the point where a model stops being a demo and starts being useful.

We keep the team deliberately small and judge a project by one thing: whether the person using it does better on their next attempt.

Hiring soonSee careers

Get early access.

We're onboarding a first group of students, and we're always glad to hear from researchers and prospective partners.

Write to us

Tell us which exam you're preparing for. One email back when we open.