MENTOR
Your AI coaching mentor
mentor.football x FIFA
FIFA AI Pro,
For The Grassroots.
Elite AI at the World Cup proved coaches will use it. This brings the same to every coach in 211 member associations: editable sessions in one click, FIFA's own coaching video, a full specialist staff, and a layer that learns each coach, team and player. Answering will commodify. This will not.
The pattern
Proven at the top.
Now for the many.
FIFA AI Pro showed that elite coaches reach for AI when it is theirs: in their workflow, grounded in their methodology, answering with their video. The value was never a smarter chatbot. It was the coaching work done for them. That same pattern brought to the grassroots is a far larger prize, not a few hundred coaches but every coach in 211 associations.
Editable in one click
Every session comes back as an editable plan you can open on the pitch, not a wall of text. Move a player, swap a drill, print it. A chatbot describes a session. This hands it to you, ready to change.
FIFA's own video, not a description
Answers show FIFA's actual coaching videos, the real footage of the technique or drill. A general model can describe an exercise. It cannot show you this clip. Yours can.
A full specialist staff
A physical coach and a psychology specialist, each with their own knowledge base, consulted separately and quoted by name. Depth per domain, not one generalist voice.
A layer that learns each coach, team and player
It remembers, over a season, who you coach and what you have worked on. The more it is used, the more it knows your context. A generic chatbot starts from zero every time, and always will.
Answering will commodify. Any model, this year or next, will answer a coaching question well. None of the above is model accuracy, and none of it commodifies. It is your content, your workflow and your data.
Own the channel. Coaches will use AI with or without FIFA. The only question is whether FIFA owns that relationship, governed, on-brand, with every conversation flowing back as insight, or cedes it to a generic chatbot that gives FIFA nothing back and carries all of the risk.
The Origin
Built from firsthand experience.
As a grassroots coach working in a multicultural, multilingual environment, I identified a clear need: coaching quality varies enormously. Not because coaches don't care, but because the best resources don't reach them.
I've been involved in football since I was seven. I've seen firsthand what happens when a coach in a remote club has no one to ask. No specialist, no curriculum, no mentor. Just instinct and hope.
I built mentor.football to close that gap, combining 40 years of football experience with 10 years in IT and 3 years building AI systems professionally.
Grassroots coach
Coaching in multicultural, multilingual environments since age 7
CS + Marketing
Master of Science in computer science and marketing
10yr IT · 15yr Digital Marketing
Global enterprise software, apps & products
3 years AI
Building production AI agents and automation systems
The Problem
Coaching quality
varies wildly.
211
FIFA member federations. One curriculum. Wildly different access to coaching support.
100+
languages spoken by grassroots coaches around the world. Most have never had a Pro-level mentor.
New
direct real-time signal on what grassroots coaches actually ask and struggle with, at scale.
The Solution
A full coaching staff.
For every coach.
Think of it as a full professional coaching staff available 24/7. Every role that elite clubs have, available to any grassroots coach anywhere in the world.
Head Mentor
Orchestrates the team, routes each question to the right specialist
Tactical Analyst
Session design, formations, game models, pressing triggers
Fitness Coach
Physical conditioning, speed development, age-specific load management
Psychologist
Motivation, confidence, behaviour management, team dynamics
Player Developer
Age-specific methodology, technical skill progression
Goalkeeper Coach
Specialist advice on demand, positioning, distribution
Architecture
Built for trust.
Not a chatbot.
Deterministic orchestration
Agents follow strict, pre-defined paths. Coaches get FIFA-approved answers every time. The system cannot hallucinate outside approved source material.
Model-agnostic
Currently OpenAI GPT-4o, but the orchestration layer runs on any infrastructure, including FIFA's preferred cloud environment.
Voice + text, 100+ languages
Built on a model that supports 100+ languages natively. A coach in any country can use it in their own language.
Offline roadmap
Hybrid approach planned: local lightweight model on-device when offline, full AI when connected. Critical for the markets FIFA cares most about.
The Bigger Picture
Unique data insight
from the grassroots.
Coaching curricula reach coaches as static documents. Until now, there has been no direct way to collect real-time data from the grassroots. Every conversation in this system is a live signal from the field.
Global skill gap map
Real-time data on what coaches in 100+ languages are actually struggling with, by age group, region, topic.
Content feedback loop
Follow-up questions signal where FIFA's curriculum needs updating or new variations.
Intent data
Knowing what coaches want to do next, not just what they have done. First time this data has existed.
GDPR-safe
Anonymised, aggregate trends only. No individual coach tracking.
The difference
Why not just
ask ChatGPT?
A general assistant will happily answer any coaching question. It will also invent a drill, guess a session length, and give the same answer to a U7 coach and a U17 coach. The difference is not the model. It is everything around it: what it is allowed to say, what it must refuse, and what it has to show its work against.
It shows the drill, not a description of one
642 drills from FIFA Training Centre, each with a diagram drawn from that drill's own coordinates and a link back to its source page. A general assistant can describe a passing exercise. It cannot show you this one.
It refuses to invent
If the bank has no drill for the theme, no diagram appears and the coach is told so. That is enforced in code, not by asking the model to behave. A missing picture is better than a wrong one.
A staff, not a single voice
A physical coach and a psychology specialist, each with their own knowledge base, consulted separately and quoted by name. Ask about a demotivated player and the psychology specialist answers, grounded in coaching literature.
Rules that cannot be talked around
No strength training under 13. Age-specific session lengths. Eating disorder and depression questions routed to health professionals, never answered. These are constraints in the system, not polite suggestions in a prompt.
Your material, not the internet average
The knowledge base is FIFA and federation material: coaching lines, load management, motivational climate. A general model averages whatever it read online, including advice your federation would never endorse.
It cannot drift from your coaching line
Age-stage limits are checked in code, not left to the model. Ask for a two-hour session for eight-year-olds and the system catches it, states the limit and tells the coach where it came from. Where your line is silent, the specialist says so instead of filling the gap with an opinion.
Every answer is checkable
Each drill links to its FIFA Training Centre page, each specialist answer names the sources it used. A coach can verify. So can you.
The risk with a general assistant is not that it fails. It is that it answers confidently and the coach cannot tell the difference. For a 9-year-old's training load, that difference matters.
Hallucination is handled by structure, not by asking the model to be careful. Diagrams come from the bank or not at all. Age-stage limits are verified in code after the answer is written, and a breach is shown to the coach with the limit and its source. Specialists cite what they used, and say plainly when your material does not cover the question. A coach carries your badge into that session, and the advice has to be one you would stand behind.
Trust at scale
Correct by construction.
Auditable by the same design.
The hard part is not the model. It is making agents a governing body can deploy to children's coaches, across 211 federations, and stand behind every answer. That is the work, and it is where most coaching AI stops.
Measured, not asserted
Correctness runs through an evaluation gate before anything ships. Every failure we find becomes a permanent test. The error rate is a number, and it only moves down.
Auditable end to end
Every answer traces to its source. Every rule check is logged. The insight numbers a federation sees can be reproduced from the underlying data, not taken on trust.
Compliant by architecture
The same structure that keeps answers correct maps to EU AI Act and ISO 42001 requirements: transparency, human oversight, data governance, record keeping. Formal classification and DPIA are part of rollout, not an afterthought.
A staff of agents, each checked
A head mentor routes each question; specialists answer in their own domain and cite what they used. The quality is not one clever prompt. It is a team of agents, each held to the same rules.
This is the moat. Anyone can wire a model to a knowledge base in a weekend. Making it measurably correct, auditable and compliant enough for a governing body to deploy at scale is a different discipline.
It is also the part that compounds. Every question the system answers, every gap it surfaces, every rule it enforces makes the next answer better and the audit trail deeper. A snapshot of the code can be copied. An operated, improving, accountable system cannot.
Under the hood
How it works.
A general model answers a question. This runs a pipeline: a specialist staff, guardrails enforced in code, grounded in FIFA's own content, returning an editable session, and getting sharper with every use.
Coach question
voice or text · 100+ languages
The engine
Orchestrator
Routes each question to the right specialist.
Specialist staff
Physical, psychology, tactical, goalkeeping. Each with its own knowledge base.
Guardrails
Age limits, safeguarding, grounding. Enforced in code, not prompts.
Grounded knowledge
FIFA curriculum, drills and video. Not the open web.
Output
grounded answer · editable session plan · FIFA video · traceable sources
The moat · feeds back ↻
Learning layer
Remembers each coach, team and player, and compounds with every use. This is what a chatbot cannot copy.
Return channel → federation
Federation insight
Anonymised and aggregated. The first real-time view of what coaches actually need.
Available now
Live today.
Scale tomorrow.
Any coach anywhere can use this right now. The question is not whether this can be built. It already exists.
Today
Public FIFA Training Centre content
Already helping coaches in 100+ languages. Live. No setup required.
With FIFA
Full curriculum + drill library
Add FIFA's complete coaching curriculum and every coach gets a pro-level mentor backed by the full weight of FIFA methodology.
Federations & clubs
Local knowledge layer on top
Each federation or club can add their own training bank and context. Same system, local identity.
The Opportunity
Pro Mentor In
Every Coach's Pocket.
Democratisation
Every grassroots coach on the planet gets the same knowledge as a Pro academy coach.
211 federations. 100+ languages. Voice or text. On any phone.
Safety
Coaches always get FIFA-approved answers. The system cannot make things up.
Deterministic orchestration layer. Responses stay strictly within approved source material. Not a chatbot.
Data insight
The first real-time signal from the grassroots, at scale, in 100+ languages.
FIFA publishes curriculum as static documents and gets no signal back. Every conversation here is a live data point from the field.
Live demo
Try it now.
Ask: "Design a 90-minute U12 session on 1v1 attacking"
Open mentor.football