The campus intelligence platform
IIT Delhi runs on.
One platform across attendance, teaching quality, accreditation, hostel operations and academics. With an AI we host ourselves, so your data never reaches an outside provider.
Deployed and in production. Run it on our cloud, or on your own campus hardware. Your choice, not ours.
Universities run on it. Companies hire from it.
The second only works because of the first.
Your institute does not have a software problem. It has an evidence problem.
You already own systems for most of this. The trouble comes when someone asks for two years of feedback data, or how an attendance figure was arrived at. The answer gets assembled by hand, by people whose time was meant for something else.
Your systems are where things get filed. Your students and faculty are somewhere else entirely.
So the record is always a reconstruction, and it is always late.
An ERP is where the institute files what happened. UAI is where it happens.
UAI is not an ERP and does not replace one. It sits in front of the institute as the place students and faculty actually spend the day, and it turns that daily use into records that were true when they were made.
Nobody is asked to file anything. The record is a by product of people going about their work, which is the only way a record ever stays accurate.
Your existing systems keep working. UAI is the layer your campus actually lives in.
Four things stop being your problem.
Attendance you can defend when someone asks
Attendance is mandated and audited, and still collected on paper by someone walking a register around a room. Ask how a figure was arrived at and there is no answer that survives the question.
The professor opens a session. The code on screen changes every few seconds, so passing a screenshot around achieves nothing. Students check in once each, on one device. The register builds itself and exports whole.
Nobody carries a register. Nothing is compiled in March.
A teaching report a professor actually wants to open
Feedback usually arrives as a number out of five, late, with no idea what to do about it. It reads as a judgement, which is why faculty resist the whole exercise.
Each professor gets a private report from their own students: what worked, what did not, what to change. Every score and chart is computed by code and reproducible by hand. The AI writes only the sentences describing those figures. It cannot produce a number, change a ranking, or reorder a priority.
The accreditation documents come out of the same data. AICTE, NAAC 1.4.1, NAAC 2.7.1 and NBA, generated instead of assembled the month before a deadline. See what a report contains.
Complaints that cannot quietly rot
Every hostel has the same failure. A complaint is raised, somebody accepts it, nobody comes, and the student has no way to say so. On paper it looks handled.
Here, accepting a job does not count as doing it. Once the time the student said they would be in has passed, they are told, and can either ask again or send it to the warden.
Staff cannot close a complaint. Only the student can.
The dean’s view shows what is still unattended after a day, and how long jobs take to accept and close, by hostel and by trade.
An AI the campus actually uses
Most institutional AI ends up as a chatbot that answers nothing and is quietly abandoned. The reason is always the same. It has access to nothing.
This one is connected to the institute itself: the course catalogue, the student’s own timetable and mail, the library, and the institute’s own documents.
And it runs on a model we host, so answering a question never sends institutional data to an outside company.
Nineteen hours a semester. Each.
We went and measured it. The walk across campus for an answer that was already published. The register going down the row while nobody learns anything. The class that moved, announced somewhere nobody was looking. The paper that existed and could not be found.
None of that is lost to bad software. It is lost in the gap between the systems and the people, which is the gap this platform exists to close.
Nineteen hours. Per person. Per semester.
Across a campus this size that is around a hundred and ninety thousand hours a term. Set against a human life rather than a budget, the institute hands back more than twenty one years of its people’s time. Every semester.
Sleipnirtech study, 2026. Method and sample available on request.
The AI is ours, not rented. Run it on our cloud or on your own campus.
Your data is not somebody else’s training set.
Almost every product now describing itself as AI powered is making a call to a model owned by someone else, in a data centre in another country, with your students’ information in the request.
UAI does not work that way. The model that answers questions is one we run ourselves, on a machine under our control, inside India. No student record, no email and no faculty document is sent to an external AI provider, because there is no external AI provider in the architecture.
One honest exception. When a question genuinely needs the open web, the search goes to a search provider, as any web search does. Institutional data does not travel with it.
Built for every university. IIT Delhi is the first to run it.
The platform is deployed and in production across IIT Delhi’s campus, not as a demonstration and not as a slide, but as what students and faculty open for the ordinary business of the term.
We would rather say that plainly than show you a wall of logos. One customer, deployed, at depth, is a different kind of evidence from a long client list, and for an institute deciding whether this will work on their own campus it is the more useful kind.
What we can show you:
Everything the institute runs on, behind one sign in.
It is listed in full because a technical evaluation should see the whole surface, not because anyone has to think about it as separate products. A student never does.
For the institute and its faculty
The work that currently lands on a department in the month before a deadline.
For the academic day
For the rest of campus life
And it works with no signal. The app holds an encrypted local copy, so a basement lab or a dead patch of campus does not mean a blank screen.
Nothing gets ripped out, and nothing waits three years.
The failure everyone has seen is the same one. A large system is bought, migration takes a year, faculty decide it is worse than what they had, and it quietly stops being used while the licence keeps being paid.
We start narrow on purpose.
One department, one semester
We begin with a single department and the modules it needs. Existing systems keep running untouched.
Students sign in with the institute email they already have
No roll list to migrate, no credentials to issue. Identity comes from the institute’s own directory.
Something visible by the end of the first semester
An attendance register with a verifiable basis, or a feedback cycle that produced its own compiled documents. Something a visiting peer team can be shown.
It widens only if it is being used
Adoption data is visible to the institute from the first week. If a department is not using it, that is an argument to stop, and we would rather you had it.
Straightforward to buy, and straightforward to justify.
We have done this work already so the department sponsoring the proposal does not have to.
We can tell you how fast a student learns. Nobody else can.
Students from our partner campuses compete every week in their chosen field. We see what they know, and we see how quickly they are getting better. You get a ranked shortlist with the evidence attached.
What institutes ask us.
What is UAI?
UAI is a campus intelligence platform built by Sleipnirtech Private Limited, New Delhi. It places a single application, a single sign in and a single AI assistant in front of everything an institute already does, and turns the daily use of that application into records the institute can rely on. It is built for every university. IIT Delhi is the first to run it, and the platform is in production on its campus.
Is UAI an ERP?
No, and it does not replace one. Existing systems for admissions, fees and examinations continue to operate. UAI is the layer students and faculty use each day, which is why the data that comes out of it reflects what actually happened.
Does UAI send our data to OpenAI or another AI provider?
No. The language model that answers questions is self hosted. It is not an API subscription to a model owned by another company. No student record, no email content, no faculty document and no question anyone asks is transmitted to an external AI provider, because the architecture contains none.
Can UAI run on our own campus?
Yes. It can be deployed on our infrastructure in India, or on hardware inside your own campus, and the product is the same in both. A server with a GPU is needed only if you want the AI running on campus.
Is UAI compliant with the DPDP Act?
We describe UAI as aligned to the DPDP framework and building toward the May 2027 deadline, not as compliance already achieved. Built today: on device biometrics, envelope encryption, guardian consent for minors, self service deletion, data minimisation, and a published privacy policy with a named grievance officer. Scheduled before the deadline: an itemised consent ledger, a data access and export facility, automated retention and erasure, and a breach notification runbook.
Which accreditation documents can UAI produce?
Compiled documents for AICTE, NAAC 1.4.1, NAAC 2.7.1 and NBA, generated from the course feedback collected during the semester rather than assembled by hand. The feedback and accreditation module is built and tested, but has not yet run through a complete semester cycle, because the feedback window opens at the end of term.
What does an institute need to provide, and how does it start?
Directory access for authentication, read access to the learning system where course material is wanted, a server with a GPU only if the AI is to run on campus, and a sponsoring department with one named contact. There is no migration and no data preparation phase. It starts with one department, one semester and chosen modules, and widens department by department on measured adoption.
Which universities can use UAI?
Any university. UAI is built for every institute, and IIT Delhi is the first to run it: the platform is in production on its campus. We would rather name the institutes we are live at than present a client list, and one deployment running at full depth is the evidence we can discuss in detail.
How can a government institute buy UAI?
Software licences are goods within GFR Rule 143, together with the installation, training and support supplied with them, so the ordinary Chapter 6 route applies. We provide a functionally written specification your department can adapt for the indent. It is contracted annually, and DPIIT startup recognition has been applied for and is not yet granted.
Start with one department and one semester.
The useful next step is a conversation about your institute rather than a demonstration of ours. Tell us which department feels the problem most, and we will show you what the first semester would look like there.
Sleipnirtech Private Limited, New Delhi. Built for every university, and IIT Delhi is the first to run it.