About EKYGAI
Our technology, our vision
1. Our vision
Sport is a domain where every detail counts: load, recovery, progression, timing. Yet most sports platforms offer generic plans or algorithms that generate text without understanding physiology.
EKYGAI was built around a conviction: the athlete, the coach and the team deserve a tool that calculates, not one that improvises. A tool that respects sports science, that adapts to each individual, and that always leaves the final decision to the human.
The platform serves three types of users:
- Athletes: personalised programme, interactive calendar, performance tracking, AI assistant.
- Coaches: supervision cockpit, programme creation, athlete and group management.
- Teams / Clubs: global oversight, multi-coach management, AI quotas, collective performance.
2. The three technological pillars
EKYGAI rests on three complementary technological layers, each with a precise role:
2.1 The DEM — Deterministic Expert Model
The DEM is EKYGAI's calculation engine. Unlike generative AI (LLMs) that produce probable text, the DEM calculates each programme with precision: the same objectives and constraints always produce the same optimal result.
The DEM operates in two layers:
- Layer 1 — General planning: analyses the objective, sequences the tasks (PERT method), structures the cycles (macro, meso, micro), selects training methods and applies safety rules.
- Layer 2 — Discipline-specific engines: five specialised engines (running, cycling, swimming, strength training, triathlon) transform abstract sessions into concrete programmes with rules specific to each sport.
Each discipline has its own duration caps, compulsory drills, volume factors and level translation tables. A beginner runner will never exceed 60 minutes in the Base phase. A beginner swimmer will dedicate 40% of their session to technique drills. These rules are encoded and tested, not randomly generated.
The DEM also incorporates five intelligent agents that execute sequentially:
- Fatigue Agent: Banister model (1975) — calculates fitness, fatigue and supercompensation.
- Safety Agent: EWMA anomaly detection — alerts if HRV drops, critical load or high stress.
- Load Module: greedy PERT algorithm — distributes tasks across weeks respecting constraints.
- Selection Module: multi-criteria scoring — chooses the best method for each session.
- Adaptation Module: re-optimises the parameters of the four preceding agents every 3 weeks.
Materialisation of each session takes milliseconds. There are no pre-filled templates: the engine recalculates every time.
2.2 The DOME — Domain Optimised Model Environment
The DOME is EKYGAI's conversational layer, embodied by EkyBot. The DOME never calculates your sessions — it relies on the DEM's outputs to explain, assist and answer your questions.
In practice, the DOME injects into each conversation a complete snapshot of your sporting context:
- Your athlete profile (EKY level, disciplines, constraints)
- Your current programme (phases, sessions, methods)
- Your current week (detail of each session)
- Your compliance over recent weeks
- Active adjustments (triggered safety rules)
- Your hierarchised objectives
EkyBot can thus answer questions such as "Why do I have LSD this week?" or "Am I progressing?" drawing on your actual data, not generalities.
The DOME is available on web and mobile, in three modes: Auto (current day context), Debrief (post-session analysis) and Weekly Review (weekly summary). In coach mode, the DOME combines the athlete's full history with the current programme.
2.3 The VAI — Vertical Artificial Intelligence
The DEM and the DOME together form what we call a VAI (Vertical Artificial Intelligence): an AI 100% dedicated to sport, built from scratch with its own concepts, methods and engines.
What distinguishes a VAI from a generic AI adapted to a domain:
- Determinism: same inputs = same outputs. No hallucination, no improvisation on your sessions.
- Specialisation: each discipline has its own engine with rules encoded by experts.
- Separation of calculation and conversation: the DEM calculates (deterministic), the DOME explains (conversational). Never the other way round.
- Sports science: Banister model, macro/meso/micro periodisation, supercompensation, EWMA — not marketing algorithms.
- Transparency: the user can ask EkyBot the reason behind every suggestion.
3. Supported disciplines
Each discipline has its own specialised engine and a verified exercise database:
- Running: injury caps by level and phase, hamstring/knee GPP, dedicated yoga.
- Cycling: progressive long rides (165min to 300min), indoor trainer (2h cap), squat/pull/core GPP.
- Swimming: compulsory drills (40% beginner to 12% elite), band warm-up, pool slot caps.
- Strength training: 609 atomic exercises, 6 muscle groups, compulsory rotation, dedicated phases (Adaptation, Volume, Intensification, Realisation).
- Triathlon: 6-step orchestration engine, distribution by discipline sliders, multi-level translation, brick bike-run.
All exercises are available in French and English. Each exercise has specific adaptations by EKY level band.
4. Governance: the human decides
Governance is at the heart of EKYGAI. It defines who takes the decisions regarding training:
- AI Autonomous: the AI suggests programmes, the athlete decides and executes. Default mode.
- Collaborative: the athlete and coach decide together, the AI provides the data and simulations.
- Full Coach: the coach directs programme creation, the AI calculates, simulates and alerts.
The athlete can change mode at any time. Even in Full Coach mode, the athlete retains the right to refuse a modification. Every governance change is logged with timestamp, actor and reason (GDPR Article 30 compliance).
5. Security and data protection
5.1 Data protection
- JWT authentication with short-lived tokens (10 min), automatically renewed.
- Passwords hashed with PBKDF2-SHA256.
- Explicit GDPR consent for each coach-athlete link.
- Immutable audit trail for governance changes.
5.2 User rights
- Right of access (Art. 15): export all your data in JSON at any time.
- Portability (Art. 20): data in a structured, machine-readable format.
- Right to erasure (Art. 17): secure deletion with email confirmation, 30-day period, full anonymisation.
- Withdrawal of consent (Art. 7): withdrawal by purpose, timestamped.
5.3 EkyBot data
- Messages are sent to the Anthropic API to generate responses. Anthropic does not retain your data beyond processing.
- Your conversations are not used to train third-party models.
- The context sent to the LLM does not contain personally identifiable data.
- You can delete, archive or export your conversations at any time.
6. The founder
EKYGAI was conceived by Redouane El Bakkouch, a pharmacist by training, elite sports coach qualified at the Sports Academy of Lausanne, and university lecturer at the Institute of Sport Sciences, Hassan I University (Settat).
Former National Technical Director (DTN) of the Royal Moroccan Triathlon Federation, he accompanied the first Moroccan athletes to qualify for the Tokyo 2020 Olympic Games and Paris 2024 Olympic Games. He also supervised the athletes who competed at the 2025 Islamic Solidarity Games, winning two medals: gold in the men's event and bronze in the women's event. Today, he remains coordinator between the Royal Moroccan Triathlon Federation and the Moroccan National Olympic Committee, responsible for high-level conventions: Youth Olympic Games, Mediterranean Games and other major international competitions.
Former national head of the windsurfing section, he is also the founder of a company specialising in water sports and has been a surfer for over 25 years.
It is this triple expertise — science (pharmacy, physiology, university teaching), the field (international competition, elite coaching) and sports entrepreneurship — that shaped EKYGAI. Every rule encoded in the DEM engine, every safety cap, every level translation table reflects years of experience with athletes at all levels, from beginner to Olympic qualifier.
7. Contact
For any questions, suggestions or demonstration requests:
- Support: support@ekygai.com
- Privacy: privacy@ekygai.com
- Registered office: EKYGAI LLP (Partnership No. OC460332), 167-169 Great Portland Street, 5th Floor, London, W1W 5PF, United Kingdom