Structuring and speeding up AI adoption at Superprof
How Superprof moved from scattered AI use to a structured, prioritised and operational approach for its marketing and product teams.
Free, no commitmentQuote within 48 hours
- business
- Private lessons marketplace
- presence
- More than 40 countries
- teams
- Marketing, product, managers
- programme
- Training, usage rules, AI Ambassadors, agents
- participation rate
- 99%
- employees trained
- 150
- satisfaction
- 4.8/5
- AI Ambassadors trained
- 12
Before, after
before
- Isolated initiatives, different tools from one team to the next
- No standard instructions
- Value hard to see, slow to reach production
after
- A common framework and shared usage rules
- Prioritised use cases, each with an owner
- Deliverables in production in weeks rather than months
Context
Superprof was already using AI, but without alignment: different tools, no standard instructions, and difficulty moving from exploration to production. The challenge: move fast without losing control. When usage grows quickly, training people, structuring the AI Ambassadors and setting the rules have to happen at the same time; otherwise adoption becomes fragile.
Objectives
- Align practices between managers and operational teams
- Prioritise use cases: impact, effort, owner, indicator
- Train AI Ambassadors able to build agents
- Deliver first results within a few weeks
Approach
A “product” method: scope, prototype, validate, document, at a pace of one day a week.
Community
Co-designed usage rules and a network of 12 AI Ambassadors: test, validate, document, share.
Marketing diagnostic
SEO, paid search, CRM, content and growth reviewed to spot gains and risks.
Build and roll-out
Prototypes, validation, then the quick wins put into production, one day a week.
Stack and governance
Dust for internal agents, Make and n8n for orchestration, Python and MCP for advanced integrations, and shared rules: instructions, security, criteria for going into production. A shared foundation avoids “shadow AI”: everyone knows where the data goes and who approves.
Results
The initiative portfolio went from 35 ideas to 14 qualified use cases, then to 5 quick wins delivered.
- Quick wins in production: automated reporting, an agent that analyses Ads campaigns, scripts and newsletters
- An operational internal community, with shared standards
- Validation criteria and limits that everyone knows
“A global, cross-disciplinary, practical AI programme, within a secure framework.”
What you can replicate in your organisation
Framework
Minimal scoping before launching anything: usage rules, data, criteria for going into production.
Community
AI Ambassadors who carry the new practices into their teams.
Cadence
Priorities reviewed every week, with an expert to deliver.