AI-powered course creation

Cracking the code of modern education is always harder when teachers are trapped in administrative loops. That was the reality for French AI startup Nolej, which entered the market to tackle a precise problem: educators and instructional designers were spending most of their working hours generating content rather than actually teaching. While their vision to automate this process was clear, their initial proof of concept needed a major upgrade. They needed a partner to help them build an enterprise-grade platform and school the competition in a rapidly developing niche.

Miquido stepped in to execute a comprehensive platform overhaul, using our AI Kickstarter framework to rebuild the frontend, backend, cloud architecture, and AI engine in just five months. Today, the platform serves 170,000 active users across 20+ countries. Emmanuel Macron even cited Nolej by name as the prime example of AI that puts teachers back in the driver’s seat.

01. Executing a full platform rebuild within a hard deadline

Nolej needed a revamped platform capable of handling significant user growth within a strict nine-month window. There was no room for a phased approach; the entire system had to move together to meet investor and partnership commitments, rather than engineering convenience. 

02. Building accurate and reliable AI capabilities

In education, content quality is non-negotiable. An AI that hallucinated or generated generic material would instantly destroy the product's credibility with teachers. The AI engine had to use creators' own source materials as the absolute ground truth. 

03. Ensuring data sovereignty for regulated markets

Nolej’s growth strategy depended on serving governmental and academic institutions across Europe. These clients have zero tolerance for data residency risks – user data and AI interactions could not leave EU jurisdiction. Building on US hyperscaler infrastructure would have closed those markets before Nolej could reach them.

04. Avoiding vendor lock-in across models and cloud providers

The AI landscape was changing faster than any single model could track. Nolej needed the flexibility to switch between LLMs mid-task and maintain infrastructure agility without making the system fragile. 

05. Developing internal AI capability within Nolej's team

Nolej didn't want a black-box system they depended on an agency to operate. Knowledge transfer had to be a core part of the delivery so their internal engineers could own the platform's future.

01. Deploying AI Kickstarter as the engineering foundation

We built the AI engine using our proprietary AI Kickstarter framework – a production-ready architecture that combines RAG pipelines, vector database integration, and multi-model orchestration. Nolej got a system built for production on day one, with tested prompt engineering, security controls, and support for the latest LLMs.

02. Implementing RAG architecture with verified source content

The retrieval-augmented generation setup connects user uploads (documents, video, audio) directly to the generation pipeline. The AI creates quizzes, flashcards, and glossaries directly from that specific material, ensuring flawless accuracy.

03. Utilizing Scaleway as the sovereign infrastructure layer

We built the platform on Scaleway to ensure data centres operated entirely within EU jurisdiction. This strategic choice bypassed US hyperscalers and unlocked the highly regulated European governmental and academic markets. 

04. Integrating multiple AI models with zero lock-in

The platform seamlessly integrates GPT-4, Mistral, Gemini, Claude, and specialized audio/video processors. Models can be switched mid-task based on performance, cost, or capability without requiring architectural changes.

05. Providing thorough knowledge transfer during delivery

Continuous workshops and real-time documentation aligned Nolej’s internal team with every architectural decision, allowing them to independently operate, extend, and scale the platform.

01. Reducing course creation time by 83%

Nolej successfully reduced lesson preparation times by 83%. A task that once took a teacher 4.5 hours now takes just 45 minutes, saving a cumulative 100,000 hours for educators worldwide.

02. Lowering content production costs by 80%

The cost of producing an hour of learning content plummeted by 80%, dropping from €225 to €37.50, allowing single educators to handle workloads that previously required full production teams.

03. Establishing European sovereign cloud infrastructure for compliance

Running cloud-agnostically on Scaleway ensured absolute data compliance under strict EU residency rules, giving the platform the flexibility to offer on-premises deployment for enterprise clients. 

04. Scaling the platform to 170,000 active users globally

The platform successfully scaled from 2,500 users to 170,000 active users globally. The automated scaling infrastructure absorbed this massive growth curve without a proportional spike in operational costs. 

05. Securing industry recognition and strategic enterprise partnerships

Nolej won the GESAwards France 2023, reached the TechCrunch Battlefield Top 20, and earned a 2024 citation by Emmanuel Macron, which helped secure critical partnerships with BPI France and EdTech France based on technical architecture credibility.