Classroom Device
Workstation, school tablet, classroom kit
Intel Core Ultra · GPU / NPU
Private AI for schools and institutions at scale.
Subscription-free, offline, Intel-powered AI platform that turns a standard workstation into shared local intelligence for four students, teachers or staff at once — no internet, no cloud reliance, and no student data leaving your campus.
- Intel Core Ultra · GPU/NPU
- 1:4 multi-user workstation
- Campus-wide, on-prem deployable
- Subscription-free
What is EdgeAI
The AI runs where the learning happens.
The same AI your students and teachers use in SchoolExl — except the brain sits inside your campus and never round-trips to a distant cloud. That’s EdgeAI.
EdgeAI Node
Intel Core Ultra · Sumeru SLMs
Answer · Vision · Voice
All locally generated
Runs on-premiseData stays localZero round-trips
The Challenge & The Alternative
Cloud AI was never built for real-world work.
Subscriptions stack up. Latency kills the moment. Sensitive data crosses borders. EdgeAI takes a different stance.
Limitations of Cloud AI
- Recurring subscriptions
- Internet dependency
- Latency
- Privacy concerns
- Limited control
EdgeAI: The Solution
- No subscriptions
- Real-time response
- 100% offline
- Full privacy
- Enhanced customization
1:4 Multi-User Workstation
One Intel Core Ultra workstation. Four users at once.
EdgeAI turns a standard workstation into a shared AI node. CPU runs normal computing; the integrated GPU and NPU run AI for up to four concurrent users — without subscription dependency or cloud reliance.
4 concurrent users
Chat & language
Audio · voice
Vision · OCR
Reasoning
Intel Core Ultra · GPU / NPU
CPU
Normal computing
GPU
AI inference
NPU
AI offload
Local SLM
Edge models
1:4
Users per workstation, simultaneously
Zero
Per-token or per-user usage metering
100%
Offline · data stays in the building
Purpose-Built SLMs
Six Small Language Models, each tuned for a job.
Sumeru is our proprietary edge-SLM suite — distilled from large LLMs to be privacy-focused, classroom-safe and on-device deployable. From 100 MB to 2 B parameters.
100 MB
Research Mini
Sumeru-rm
The lightest, fastest model — tuned for short queries and embedded workflows.
256 MB
Audio
Sumeru-a
Audio-native: voice interaction, sound identification, real-time transcription.
256 MB
Thinking
Sumeru-t
Step-by-step reasoning for plans, breakdowns and structured analysis.
450 MB
Vision
Sumeru-v
Vision-based intelligence — diagram parsing, OCR, handwritten understanding.
1 B
Intelligent
Sumeru-i
The everyday workhorse. General language understanding across domains.
2 B
Research
Sumeru-r
Deep research and long-context analysis for knowledge-heavy workflows.
Diverse Capabilities
A domain-ready foundation for any workflow.
Five core capabilities — combined and routed across the Sumeru SLMs on the workstation.
Language understanding
Read, summarise, draft and answer across domains — locally, with full context.
Audio & voice interaction
Hands-free Q&A, dictation and natural conversation — no wake-word lag.
Sound identification
Detect events, faults and anomalies from raw audio — across a campus, on the factory floor or in the ward.
Vision-based intelligence
Read a whiteboard, scan a worksheet, mark handwritten work, interpret a diagram — locally.
Offline reasoning
Plan, decide and explain — even on workstations with zero internet connectivity.
Total cost of ownership
Cut your LLM bill by 60–80% over five years.
Cloud-based LLM APIs charge per token, every month, forever. EdgeAI is a one-time depreciable asset that gets cheaper every day it runs.
1:4
Concurrency
Users per workstation, simultaneously
Zero
Per-token cost
No usage-based charges, ever
0
SLMs
Audio · Vision · Thinking · Intelligent · Research
0%
On-prem
Inference never leaves your building
The 5-year view
Over five years, recurring cloud LLM bills compound month after month. EdgeAI replaces them with a one-time depreciable asset and zero recurring inference cost — institutions typically reclaim 24+ months of cloud spend.
Ideal Use Cases
Built for any place where data can’t leave the building.
Education
Classroom-safe AI for schools, colleges and labs — curriculum-aligned assistants, assessments and tutoring for every student, with zero student data leaving campus.
Branch offices
Offline field operations for banks, government and logistics. Run the same AI assistant whether the link is up or not.
Healthcare
Patient-data-safe assistants for clinical notes, imaging triage and ward operations — never leaves the building.
Manufacturing
On-line, on-floor inference for defect detection, work-instruction copilots and sound-based fault alerts.
Retail
Store-side assistants for inventory, planogram checks and customer Q&A — even when the back-of-house WiFi drops.
Secure enterprise
Knowledge assistants for legal, defence and regulated industries. Air-gapped by design, audited by default.
Why Institutions Care
Four reasons EdgeAI ships into production.
No subscription cost
One-time deployment on hardware your institution owns. No usage metering, no recurring seat bills. The unit economics finally make sense at campus scale.
Offline by design
Internet-independent. Student and staff data stays inside the building. Compliance teams sleep easier; latency is predictable.
Performance-efficient
Inference is offloaded to the integrated GPU/NPU, leaving the CPU free for normal computing.
Low-latency local inference
Local node inference layer — responses in milliseconds, not seconds. No round-trips to a distant region.
Security & compliance
Built for the room where compliance lives.
EdgeAI is what you hand to the IT lead or CISO who has stopped trusting “trust us” — six audit-ready pillars, zero cloud round-trips.
Data residency
Every byte of inference and storage stays inside your network perimeter. No data leaves the building.
Air-gap deployment
Run EdgeAI completely disconnected from the internet. Updates delivered via signed offline packages.
FERPA · COPPA · GDPR ready
Student-data compliance documentation, SOC2 evidence, DPA templates and audit packs for your legal review.
Audit logs
Every query, every response, every model call — captured, signed, exportable to your SIEM.
Encryption everywhere
AES-256 at rest, TLS 1.3 in transit. Hardware-secured key storage on the workstation.
Role-based access + SSO
RBAC out of the box, SSO/SAML integration with Okta, Azure AD and Google Workspace.
Compliance packs available on requestDPA · SOC2 · ISO 27001 mapping · pen-test reports
Proven & Extendable
Already shipping. Ready to extend to your workflow.
Current Implementation · SchoolExl
Built for classrooms.
EdgeAI already powers SchoolExl — the World’s first Safest, Smartest and Curriculum-Aligned Agentic AI for Education. Two production modules ship today, with more on the way.
- Curriculum assistant
- Assessments engine
Enterprise-Ready
Built for your business workflow.
The same platform extends to legal, healthcare, manufacturing, banking — anywhere data has to stay local and the workflow has to run.
- Admin layer
- Custom dataset training
- Fine-tuning
- RAG on private data
- Workflow adaptation
Deployment paths
Three ways institutions deploy EdgeAI.
Start with one workstation in a single lab. Scale to a campus, a district or a whole learning network. Same SLMs, same security posture — we scope the rest with your procurement team.
Single Workstation
One workstation for a single lab, staff room or department — the usual starting point for an onsite pilot.
- 1 Intel Core Ultra workstation
- 1:4 concurrent users
- Audio + Vision + Intelligent SLMs
- Basic admin dashboard
- Email support (next-business-day)
Most deployed
Campus
5–25 workstations across a school, college or campus — labs, staff rooms and the assessment cell.
- 2–25 workstations
- All 6 Sumeru SLMs
- Centralised admin + SSO
- Custom dataset training on your curriculum
- Priority support · 4-hour SLA
District & Network
Multi-site districts, learning networks and regulated institutions, with custom fine-tuning.
- Unlimited workstations
- On-prem air-gap deployment
- Custom model fine-tuning
- Dedicated solutions architect
- Premium support · 1-hour SLA + 24/7 escalation
MSA · DPA · tender documentsStudent-data review pack prefilledPO · NET-30 · NET-60 accepted
Every deployment is one-time hardware your institution owns plus an annual support contract, scoped with your procurement team. No usage metering, no surprise bills.
Procurement, answered
Eight questions school IT and procurement always ask.
An Intel Core Ultra workstation with integrated GPU and NPU. We ship pre-configured units to your campus, or your IT team can provision from a hardware spec sheet we provide. One workstation serves four concurrent users — a lab pod, a staff room or an assessment cell.
Have a question not covered here? Email edgeai@schoolexl.com — your reply lands in our solutions architect's inbox, not a ticket queue.
The Vision
The classroom of tomorrow runs on EdgeAI.
Private. On-premise. Subscription-free. For every learner, every district — from a one-room school in Bihar to a 5000-student campus in Singapore.
Every Learner
AI tutors that know every learner by name.
From the first day of class to the final mock, on-device.
Every Classroom
Curriculum-aligned, teacher-trained.
Built with educators, audited by them.
Every District
On-prem at scale, zero subscription drag.
One-time deployment, year-after-year compounding.
Subscription-free, offline, Intel-powered AI for every campus.
Plan a deployment with our solutions team. We’ll size the workstations, shortlist the SLMs and map them to your campus network in a single call — measurable impact from Week 1.