AI Photo-Attendance, Running On-Premise Inside a University
Nexgale took its own AI attendance product, ClassSight, and rebuilt it to run entirely inside Parul University's private network — the same photo-based attendance, but with student data that never leaves the university's servers and records that flow straight into the systems staff already work in.
The challenge
Parul University wanted what ClassSight does: a teacher photographs the class, the AI matches enrolled students, and faculty confirms the roll — replacing manual roll-calls. But the standard cloud version couldn't meet the university's requirements. Student data had to stay on the university's own servers, the system had to run on the university's own infrastructure, and attendance had to land directly in the MIS the institution already relies on — not in a separate tool.
These weren't preferences to negotiate around: for an institution holding student records at scale, where data lives and what systems it touches are first-order constraints. A general-purpose hosted product, however capable, was the wrong shape for the problem.
Our approach
Rather than ask the university to bend to the product, Nexgale bent the product to the university. We built a custom, on-premise version of ClassSight that keeps the same AI attendance capability but deploys inside Parul's private network and integrates with their internal MIS.
This is the work Nexgale is built for: designed, built, and operated to run in the client's stack, with their data staying theirs. The AI matching that powers ClassSight stayed intact — what changed was where it lives and what it talks to: moved onto the university's own infrastructure and connected to the records system already in daily use.
What we built
AI photo-attendance, retained in full
The core ClassSight workflow carried over unchanged: a teacher photographs the class, the AI matches against enrolled students, and faculty confirms the roll. No capability was traded away to make it run privately.
On-premise deployment
The system runs on the university's own infrastructure, inside their private network, so the attendance workload lives on hardware the institution controls rather than on external cloud servers.
Student data that stays on the university's servers
All student data remains on Parul's own servers. Nothing about the matching or record-keeping requires that data to leave the institution's environment — by design, the data stays theirs.
Direct integration with the existing MIS
Attendance flows straight into the management information system the university already uses, so records land where staff expect them instead of in a separate, parallel tool that would need reconciling.
Outcome
ClassSight is built, tested, and live on Parul's own infrastructure — now rolling out across the university department by department. It's already running in the first 4–5 departments, covering roughly 500 to 1,000 students, with the rest being brought on one at a time.
In every department that's live, attendance happens inside the records system staff already use — no separate tool to reconcile, and student data that never leaves the university's servers. That's what "runs in your stack, your data stays yours" looks like in practice: a product capability reshaped into a private deployment that respects an institution's real data and infrastructure boundaries, without diluting what it does.
Full rollout is ongoing — we'll expand this case study as it completes.
What it runs on
Have a problem shaped like this?
Tell us the work slowing your team down. We design, build, and run the AI that handles it — in your stack, with your data staying yours.