Merzal AI
Pilot deployment scheduled

Merzal AI
Private by design.

A private AI platform for educational institutions — built so your students, faculty and administration can use AI without your institution's knowledge leaving your control.

Book a demo Request a pilot
Launching at
Mailam Engineering College
Department
Artificial Intelligence & Data Science
Status
Deployment begins next month

AI is already on your campus. None of it belongs to you.

Faculty are paying for consumer subscriptions out of their own pockets. Students use whatever is free that week. Somewhere between the two, a policy draft, a student record or an accreditation file gets pasted into a chat window — and leaves the institution for good.

None of that is anyone's fault. The tools are genuinely useful, and no one gave people an alternative. But it leaves management unable to answer a question they will eventually be asked: where does our institutional knowledge live, and who is accountable for the answers it produces?

Merzal AI is the alternative. One platform, deployed inside a boundary you draw — on your own hardware if you want it there — grounded in your syllabus, your circulars and your records. Students get an assistant that knows your institution. Staff get agents that do the paperwork. Management gets an audit trail. And the knowledge stays where it started.

Grounded
Every answer cites your own documents
Owned
Your index, your data, your hardware
Role-aware
Chairman to student, scoped by design
Deployable
On-premise, hybrid or private cloud

What it does

Four things it does on day one

Every module runs under one administration and one security model, so your IT team manages a single platform rather than eleven subscriptions.

Student Assistant

Answers from your syllabus, your regulations and your past question papers — not from the open internet. Available every hour of the term, including the night before an exam.

Staff Automation Agents

Circulars drafted, reports compiled, accreditation evidence assembled. The agent does the first pass and hands it back for approval; a human still signs.

Institutional Knowledge Base

One retrieval layer over every circular, syllabus and record you hold — scoped so a department sees its own material and nothing it shouldn't.

Document Intelligence

Ask a question across a two-hundred-page regulation and get the answer with the page cited, so the reader can check the source rather than trust the summary.

And four more, once the first four are bedded in

OCR Ingestion

Printed circulars, scanned records and handwritten registers become searchable institutional knowledge.

Voice AI

Hands-free queries and dictation for faculty and office staff, in the languages your campus actually speaks.

Management Analytics

Adoption, query themes and workload patterns — evidence for management review, not vanity charts.

Role-Based Access

Chairman, principal, dean, HoD, faculty and student roles, with department isolation enforced by design.

Explore the full platform

Where it runs

You decide where the boundary is drawn

The same platform, three control models. Start wherever your governance is comfortable and move later without retraining a single user.

Private Cloud

A dedicated tenancy in India. The fastest route to a live pilot, with a clean migration path on-premise when you're ready.

  • No shared infrastructure
  • Managed patching and monitoring
  • Live in weeks

Hybrid

Knowledge, documents and retrieval stay on campus. Elastic capacity absorbs the peaks around admissions and results.

  • Index stays on campus
  • Burst capacity on demand
  • Recommended for most colleges

On-Premise

The entire platform runs on institution-owned hardware, inside your own network perimeter. Nothing routes outward.

  • Nothing leaves the campus
  • Institution owns the stack
  • Full offline operation

Underneath, four layers

  1. 01

    Access & Roles

    Single sign-on, role mapping and department scoping applied to every request before it is answered.

  2. 02

    Institutional Memory

    Ingestion, OCR, structuring and versioned retrieval across the documents you already hold.

  3. 03

    Reasoning & Agents

    Private inference and task agents that draft, compile and prepare work for a human to approve.

  4. 04

    Governance

    Audit logs, answer attribution, retention rules and the analytics management reviews.

Every layer is auditable, and institutional data never has to cross outward for a model to be useful.

See the deployment architecture

Why the boundary holds

Privacy is an architecture decision, not a clause in a contract

Any vendor can promise confidentiality in writing. Only architecture can demonstrate it. Merzal AI keeps inference, retrieval and storage inside the boundary your institution governs — which means the promise is checkable rather than contractual.

  • Private infrastructure — no shared multi-tenant model
  • The institution owns its index, documents and embeddings
  • Encrypted in storage and in transit
  • Zero training on institutional data
  • Department and document-level isolation
  • A full audit log of every query and agent action
Read the security model

Compared honestly

Where a general assistant wins, and where it can't help you

ChatGPT and Copilot are better than us at general knowledge, and we are not going to pretend otherwise. The difference shows up on institutional work, where the architecture decides the outcome.

  • Where institutional data lives

    Inside your campus or your private cloud tenancy

    Compare with the alternatives
    ChatGPT Team
    Vendor cloud, outside India in most cases
    ChatGPT / Claude Enterprise
    Vendor cloud, contractual controls only
    Microsoft Copilot
    Microsoft tenant, tied to M365 licensing
    LMS chatbots
    Third-party SaaS with no data boundary

  • Ownership of knowledge base

    Institution owns the index, documents and embeddings

    Compare with the alternatives
    ChatGPT Team
    Uploads live in vendor workspaces
    ChatGPT / Claude Enterprise
    Vendor-hosted, exportable at best
    Microsoft Copilot
    Bound to Microsoft Graph
    LMS chatbots
    Vendor-owned

  • Grounded in your syllabus & circulars

    Yes — department-scoped retrieval over your own records

    Compare with the alternatives
    ChatGPT Team
    Manual per-user uploads
    ChatGPT / Claude Enterprise
    Possible with heavy internal build
    Microsoft Copilot
    Only what already sits in M365
    LMS chatbots
    Scripted FAQ answers

  • Staff automation

    Agents that draft, file and follow up on routine work

    Compare with the alternatives
    ChatGPT Team
    Chat only
    ChatGPT / Claude Enterprise
    Requires you to build and host agents
    Microsoft Copilot
    Document assistance inside Office
    LMS chatbots
    None

  • Cost model

    Institution-wide platform fee, no per-seat charge

    Compare with the alternatives
    ChatGPT Team
    Per seat, every month, forever
    ChatGPT / Claude Enterprise
    Per seat with an annual minimum
    Microsoft Copilot
    Per seat on top of M365
    LMS chatbots
    Per conversation or per seat

  • Role-based administration

    Chairman, principal, HoD, faculty and student roles built in

    Compare with the alternatives
    ChatGPT Team
    Flat workspace roles
    ChatGPT / Claude Enterprise
    Admin console, not education-aware
    Microsoft Copilot
    Tenant-level, IT-managed
    LMS chatbots
    Minimal

  • Audit & accountability

    Attributable answers with source citation and full audit log

    Compare with the alternatives
    ChatGPT Team
    Limited
    ChatGPT / Claude Enterprise
    Log export available
    Microsoft Copilot
    Tenant audit logs
    LMS chatbots
    None

  • Deployment choice

    On-premise, hybrid or private cloud

    Compare with the alternatives
    ChatGPT Team
    Vendor cloud only
    ChatGPT / Claude Enterprise
    Vendor cloud only
    Microsoft Copilot
    Microsoft cloud only
    LMS chatbots
    Vendor cloud only

  • Support model

    Named engineers who deploy inside your campus

    Compare with the alternatives
    ChatGPT Team
    Self-serve documentation
    ChatGPT / Claude Enterprise
    Tiered enterprise support
    Microsoft Copilot
    Partner-led
    LMS chatbots
    Ticket queue

What to expect

What changes in the first two terms

These are projections from scoped department deployments, not measured results from a live campus — we have one pilot beginning and we will report against these numbers publicly, quarter by quarter.

0%

Administrative hours reclaimed per faculty member

0/7

Grounded student support across the academic year

0%

Institutional knowledge inside your control boundary

0 platform

Replacing scattered per-seat subscriptions

What it costs

Priced per enrolled student, not per seat

On-premise licensing is ₹1,200–1,500 per enrolled student per year, and that figure covers faculty, staff and every module. Set your institution's size below and see the number.

Configure your institution

Model your AI cost before you commit

Priced per enrolled student — faculty, staff and every module included. Planning estimates, not a quote.

Enrolled students
2,500

The only figure you are billed on.

Faculty & staff
180

Included at no additional cost

Compare against

Per-seat pricing is costed for all 2,680 people on campus, because that is the coverage Merzal AI gives you. Licensing a smaller group is a cheaper bill for a smaller platform, not the same thing bought two ways.

Your projection

Per-seat licensing

$804,000

per year · 2,680 seats at $25/user/month

Grows with every person you add, and stops the day you stop paying.

Merzal AI · institution-wide

94%

Lower than per-seat licensing

$759,000 kept every year — with all 2,680 people on campus still covered.

$1.50 per student / month · $45,000 a year

How this differs

What you pay for

Per-seat: A paid seat for each of 2,680

Merzal AI: 2,500 students — staff included free

Cost as you grow

Per-seat: Rises with every seat added

Merzal AI: Flat, per enrolled student

Where your data lives

Per-seat: The vendor's infrastructure

Merzal AI: Inside your campus network

If the contract ends

Per-seat: Access stops, nothing remains

Merzal AI: You keep the index and documents

On-premise deployment

The platform runs inside your network perimeter. Nothing leaves the campus, and the institution owns its index, documents and embeddings.

Comparison uses published list pricing and excludes education discounts, which vary by institution. No institution licenses AI campus-wide today — this models what matching Merzal AI's coverage would cost. Server sizing is scoped separately.

Get a scoped quote

Pilot programme

The first deployment is scheduled. The next slots are open.

Merzal AI enters live deployment next month at Mailam Engineering College, in the Artificial Intelligence & Data Science department. We are opening a limited number of pilot slots to institutions in Tamil Nadu this academic year, and a department pilot carries no licence fee.

Request a pilot slot
Deployment scheduled

Mailam Engineering College

AI & Data Science Department

Scope
Department pilot
Deployment
Hybrid
Users
Faculty & students
Begins
Next month

Built to expand to multiple institutions from day one.

Questions

The questions management asks first

Direct answers. If something here is missing, ask us on the call.

Bring a private AI platform onto your campus this academic year.

Thirty minutes with your management team, then a scoped pilot for one department. No licence commitment to begin.

Book a demo Request a pilot
Book Demo ROI