Gen3Block
Start here
Gen3Block
Sign inStart here
All case studies
EdTech

IUFP

International Education · 4 months, scoping to production

5x Faster Processing Across 10+ Countries

A Virtual Learning Environment, automated processing pipeline and AI admissions assistant built for an international EdTech provider operating across more than ten countries.

  • 5x reduction in administrative processing time end to end
  • 10+ partner schools operational across 6+ countries within 4 months
  • Student queries answered instantly, 24/7, down from a 24 to 48 hour admin turnaround
5x
faster processing
Instant
replies, down from 24 to 48 hrs
4 months
to production
Visit iufp.org.uk
IUFP foundation programme students in branded IUFP t-shirts, standing together outdoors

IUFP Students

International University Foundation Programme

The situation

The starting point

The International University Foundation Programme coordinated university-entry foundation courses across more than ten partner schools in six countries. Every application, partner onboarding, and document exchange was managed through spreadsheets and email. Processing delays were routine, data was inconsistent between schools, and scaling to new geographies was practically impossible without hiring more staff.

  • Partner school onboarding required weeks of manual coordination and document exchange via email
  • Student applications had no structured status tracking: progress was invisible to admissions teams
  • No data isolation between partner schools, creating compliance and confidentiality risks
  • Six-plus countries operating on different systems, timelines, and access requirements with no central view
  • Document uploads, validation, and delivery all handled manually with no audit trail

What was delivered

What the Platform Does

A multi-tenant education management system that brings every school, every student, and every partner into a single controlled environment, with access enforced by role at every level.

  1. Isolated Partner School Portals

    Each partner school operates in its own walled environment. Resources, student data, and credentials are fully isolated by design. No partner can see or access another school's operations.

  2. Structured Application Pipeline

    Every student application moves through a defined workflow: pending, under review, accepted, or rejected. Admissions teams always know exactly where each applicant stands.

  3. Document Management

    Students submit documents directly through the platform. Files are validated at upload, stored securely, and delivered to the right admissions handler automatically. No email attachments.

  4. Automated Partner Onboarding

    New schools apply through a self-serve flow. Once an admin approves, the school's portal and login credentials are provisioned automatically. No IT ticket, no waiting, no manual setup.

  5. 6-Tier Access Control

    From Student to Super Admin, every role gets exactly the access it needs and nothing more. Least-privilege enforced across the entire platform at every level.

  6. Full Audit Log

    Every login, document access, and admin action is recorded. Session management with force-logout capability built in for security, compliance, and peace of mind.

The outcome

The Outcome

A platform that replaced a spreadsheet and email operation with a scalable, compliant system, delivered in four months and operational across six countries from day one.

  • 5x reduction in administrative processing time across the full admissions pipeline
  • 10+ partner schools operational with isolated access within 4 months
  • Automated partner onboarding: new schools go from application to operational without manual IT work
  • GDPR-compliant data flows with full audit trail throughout
  • Real-time visibility into application status, partner operations, and student progress
  • Student admissions and visa queries answered instantly, 24/7, down from a 24 to 48 hour admin turnaround, with every answer linked to the IUFP guidance it came from

Second build

An AI Admissions and Visa Assistant

Alongside the VLE, Gen3Block built IUFP a conversational assistant that answers student questions on UK university applications and student visas. It is a retrieval-augmented generation (RAG) system: every answer is drawn from IUFP's own published guidance rather than the model's general knowledge, so advice stays accurate, consistent and on policy. Students use a web chat interface backed by a FastAPI service and a Postgres vector database.

  1. Knowledge base

    A Single Source of Truth

    Answers come only from IUFP's own admissions and visa guides, held in secure cloud storage. The assistant never draws on the open web, and each answer links back to the source document, so students can verify it and every response reflects the position IUFP has approved.

  2. Ingestion

    Incremental, Previewable Updates

    Guides are split into overlapping passages and embedded into the vector database. Only new or edited documents are processed, edits replace earlier versions atomically, withdrawn documents are pruned, and every run can be previewed with a change list and cost estimate before it touches production.

  3. Retrieval

    Hybrid Semantic and Keyword Search

    Each question runs vector similarity search and full-text search in parallel, blended into one weighted score. Semantic search catches questions phrased in the student's own words; keyword search catches exact terms such as visa routes and document names.

  4. Grounding

    A Relevance Gate Before Generation

    Only passages that clear a calibrated relevance threshold reach the language model. When nothing in the guidance is relevant enough, the assistant says so rather than guessing.

  5. Answer policy

    No Invented Figures

    The model answers strictly from the retrieved guidance. It is instructed never to invent or calculate figures such as fees or deadlines, and to state plainly when a question sits outside the guides. Low-temperature generation keeps answers consistent from one student to the next.

  6. Operations

    Hardened and Right-Sized

    Per-client rate limiting, admin-key protection on usage statistics and conversation history, health checks that confirm database connectivity, and validated connection pooling. The database is reachable only over a private network, and infrastructure is sized to real demand so running costs stay proportionate.

“The IUFP VLE has completely transformed how we manage our global academic network. We now have full visibility of student applications, admissions, and partner operations, all within a single, intelligent system.”

Bola Makinde

CEO, IUFP

All case studiesHow we approach AI implementation →