LIVE BETA · ultra.zenx.asia

Intelligence for
Endurance.

ZenX Ultra is not a training app.
It is a human performance intelligence system
built for ultra trail runners who operate
under extreme conditions.

OPEN APP → VIEW ROADMAP
SCROLL
100+ TRAIL RUNNERS
BETA COMMUNITY
11 INTELLIGENCE ENGINES
IN PRODUCTION
3+ TRAIL COMMUNITIES
CCTC · TRAIL STATION
0 FAKE ENGINE OUTPUTS
EVERYTHING COMPUTED

THE PROBLEM

Most athletes DNF
not from weakness —
from lack of intelligence.

REAL DATA · DALAT ULTRA TRAIL 2026 · FINISH RATES
50KM
89%
787 / 881 finishers
75KM
44%
140 / 315 finishers
100KM
62%
182 / 292 finishers
100KM PRO
50%
13 / 27 finishers
Source: Dalat Ultra Trail official data · 2026 · Temperature swing 14–36°C
🌡
Environment not modeled
Temperature swings of 14–36°C in a single race. Most athletes train in controlled conditions and arrive unprepared for thermal stress. ZenX Environment Engine models this automatically.
🧠
Mental load overload
"Thứ giết bạn trước không phải đôi chân, mà là cái đầu quá tải." Athletes without a systematic plan collapse cognitively before physically. ZenX Behavior Engine detects this pattern early.
📉
Race severity underestimated
Athletes register for races beyond their current adaptation capacity. ZenX Race Severity Index (RSI) quantifies this mismatch — before race day, not during.
"

Nguyên nhân DNF 100% là do người mua bib, người không mua bib thì không sợ DNF. Tỷ lệ "gãy" nhiều nhất do chênh lệch nhiệt độ 14–36 độ. Sau 20 tiếng trong rừng, đồng hồ Suunto Vertical ghi nhận hơn 12,000 calo... Trong race thường sẽ nói "lần sau đừng đứa nào rủ tao đi 100km nữa" — nhưng khi nghe giá SEB sale 50% là lại mua để dành nửa năm tính tiếp.

🏃
Phạm Ngọc Đức
Co-founder CEO Runner Club · Dalat Ultra Trail 100km · đồng hành ZenX
A 50% DNF rate at 100km Pro is not an athlete problem. It is an intelligence gap problem.
ZenX was built to close it — with data, prediction, and adaptive coaching.

POSITIONING

Not a dashboard.
An intelligence system.

Most apps show you data. ZenX Ultra models your behavior, predicts your performance, and adapts its recommendations — automatically, based on your real physiological and training signals. Every output is explainable. Every recommendation cites its source.

Fitness tracker
Performance intelligence
AI wrapper / chatbot
Deterministic engine stack
Fintech / sporttech app
Endurance intelligence platform
Generic recommendations
Your data · Your numbers · Your plan

INTELLIGENCE ARCHITECTURE

13 engines.
One intelligence loop.

Signal → Model → Context → Predict → Adapt → Feedback → Optimize.
Every engine feeds the next. Nothing is isolated.

SIGNAL
CONTEXT
MODEL
EXECUTION
FEEDBACK
AI
LIVE
IN PROGRESS
SIGNAL
2 engines
Athlete Profile Engine
Builds the athlete's physiological baseline — adaptation capacity, fatigue tolerance, injury risk, and training age — from onboarding data and race history.
Training Signal Engine
Ingests weekly training data from Strava and manual logs. Computes readiness, load density, vertical stress, and fueling consistency scores.
CONTEXT
2 engines
Environment Engine
Models race-day weather via OpenWeather. Computes thermal stress index and hydration demand multiplier. Auto-feeds RSI and Prediction Engine.
Race Model Engine — RSI
Computes Race Severity Index from distance, elevation gain, terrain technicality, cutoff pressure, and checkpoint complexity. Quantifies race difficulty on a 0–100 scale.
MODEL
3 engines
Prediction Engine
Estimates finish probability, predicted time, and cutoff safety margin. Generates per-checkpoint arrival trajectory. Every prediction is an immutable, traceable snapshot.
Fueling Engine
Generates per-checkpoint fueling strategy with caloric and fluid targets. Accounts for real CP km marks, athlete metabolic profile, and total race duration.
Qualification Engine
Models race qualification requirements across UTMB, ITRA, VTS, URS, and AIM. Qualifications are race-level properties — extensible without schema changes.
EXECUTION
3 engines
Plan Engine
Authors the weekly training plan and tracks execution. Matches planned sessions against Strava actuals. Computes adherence, today status, and upcoming sessions.
Checklist Engine
Generates race-specific mandatory gear lists from the registry. Supports DSQ, Time penalty, Weather-conditional, and Recommended buckets with official race org notes.
Gear Engine
Tracks shoe and trail pole lifecycle by terrain type. Scores terrain match and fit. Alerts before race-day gear failure. Integrated with the mandatory gear checklist.
FEEDBACK
3 engines
Post-Race Engine
Captures race outcomes and compares against pre-race predictions. Feeds deviation data back into the intelligence loop. Triggers AI-generated recovery and re-planning.
Behavior Engine
Detects overtraining patterns, load spikes, and recovery neglect across time. Requires 3–4 weeks of real training signal data to activate. Currently collecting.
Season Engine
Models multi-race periodization: base → build → peak → taper → recovery. Detects race conflicts and season readiness. Requires 2+ registered races to activate.
🧠 AI Coach — adviser layer only. Claude Sonnet receives all engine outputs as context and provides explanations and recommendations. AI may NOT override deterministic engine outputs. Every recommendation cites its source signal. This is a fundamental ZenX architecture rule.

DEVELOPMENT JOURNEY

From zero
to intelligence.

Every phase is architecture-driven. No feature is built before its engine foundation exists.

D0
Architecture Foundation ✓ DONE · May 2026
Canonical Architecture 8 Engines Defined AI Governance Model Tech Stack Selected
D1
Core Engine Implementation ✓ DONE · Mar–May 2026
Athlete Profile Engine Signal Engine RSI Race Engine Prediction Engine Gear Engine AI Coach v1 Firebase · Auth · i18n Live on web.app
D2
Intelligence Expansion ✓ DONE · May 2026
Environment Engine AI Coach Fix i18n VI/EN Complete Full UI Polish · 5 pages CoachPage v2 · PLAN tab Post-Race Intelligence
D3
Pre-Race Intelligence ✓ DONE · May 2026
Race DB v3.0 · Sub-collection schema RSI Engine v3 · Checkpoints Prediction Engine v3 · Per-CP trajectory Checklist Engine v3 · Registry-backed Admin Topo Atlas Design System Race Data Quality Panel · DataLog Admin Dashboard v2 · 11 widgets Config System · Firestore-backed
D4
Integration & Intelligence Maturity ✓ DONE · Jun 2026
Strava OAuth · Activity Sync sessionSignals Pipeline Provider Provenance · AI Firewall Admin Integrations Config Training Pipeline Fix Garmin / COROS Adapters ⚠ approval pending
D5
Coach × Training Loop · Engine Extensions ● ACTIVE · Now
Coach Plan authoring · planItems[] schema Training Execution Dashboard Weekly Plan Eligibility Gate Plan Engine · adherence tracking Training Block Engine (Base→Taper) APF Engine · Athlete Proficiency Tier Offline Race Day Mode Subscription "Tô Phở" ZenX Score · ELO 0-1000
D6
Network Effects UPCOMING
ZenX Score Public Community Cohort Benchmark Copy Ultra · Elite Coach Base ZenX Trade Alpha
D7
Platform Monetization · Governed Cognition FUTURE
ZenX Core Monorepo Multi-domain Platform Elite Monetization Network Autonomous Intelligence Layer

STRATEGIC MOAT

Why ZenX
is hard to copy.

01
BEHAVIOR INTELLIGENCE
We model the human — not just the workout. Behavioral patterns, recovery neglect, overtraining signals. This accumulates over time and becomes deeply personal.
02
PREDICTIVE MODELING
Finish probability, cutoff safety margin, fatigue trajectory — computed from your data, not averages. Every prediction is traceable to its input signals.
03
ADAPTIVE OPTIMIZATION
The system adapts week by week. Environment, race severity, athlete history — all feed into dynamic recommendations that evolve with the athlete.
04
EXPLAINABLE AI
AI never generates opaque outputs. Every recommendation cites its source engine signal. Deterministic-first, AI-advised. Athletes trust what they understand.

FOUNDER STORY

Built by a runner
who needed this.

ZenX Ultra was not built from a boardroom. It was built from 250km of mountains, dozens of hours alone on trails, and a hard-earned realization:

"Thứ giết bạn trước thường không phải đôi chân, mà là cái đầu quá tải. Mệt không đáng sợ. Hoảng mới đáng sợ."

— Hà Thanh Phong · Founder, ZenX Ultra

After races spanning 2019 to 2025 — from his first 45km to back-to-back 100km finishes — he saw the same pattern: athletes with perfect training logs making wrong decisions in the field — because they had no system to interpret their own data.

ZenX is his answer. Built to get him to 100 Miles in 2026, and 250km in 2027. Not a dashboard. An intelligence system that thinks with you — so you can think less, and go further.

FOUNDER RACE RECORD · VERIFIED
Prenn Trail Summit 100km
Dec 2024 · 106.0km · 4,053m D+
22:09:00
🏆 TOP 1 · Age Group
Dalat Ultra Trail 100km
Mar 2025 · 100.19km · 4,415m D+
22:19:06
Brăh Yàng Trail Summit 60km
Mar 2025 · 60.96km · 3,847m D+
14:50:00
ALSO COMPLETED
CCTC Challenge II · 55.63km · 5,168m D+ 20:57:48
CCC Season 4 · 55.18km · 5,007m D+ 20:34:04
DLUT 2024 · 55.16km · 2,641m D+ 10:05:09
DLUT 2023 · 25km 3:32:48
DLUT 2019 · 45km · first race 7:39:25
🎯 NEXT TARGETS
100 Miles Prenn Trail Summit 2026
100km Brăh Yàng Trail Summit 2027
250km Prenn Trail Summit 2027
Building ZenX to get there. Skin in the game.
THE PHILOSOPHY BEHIND ZENX — IN THE FOUNDER'S OWN WORDS

"Running an ultra trail is not just about covering the distance,
but conquering yourself with every step."

— Hà Thanh Phong · Founder, ZenX Ultra
STILLNESS IN MOTION

"Muốn đi xa hơn, phải nghĩ ít hơn. Cái gì không giúp mình tiến về phía trước, bỏ."

SYSTEM OVER EMOTION

"Người chạy ultra giỏi không phải là người chịu đau tốt hơn — mà là người quản lý được sự hỗn loạn trong đầu tốt hơn."

MEASURE → PREDICT → ADAPT

"Không phải 250km. Mà là từ đây tới checkpoint tiếp theo. Cứ vậy mà đi. Từng đoạn nhỏ."

INSPIRED BY · HARUKI MURAKAMI
"What I Talk About When I Talk About Running" · 2007

"Pain is inevitable. Suffering is optional."

→ ZenX Behavior Engine detects when suffering becomes data

"Most runners run not because they want to live longer, but because they want to live life to the fullest."

→ ZenX exists to protect that fullness — with intelligence

"My opponent is myself of yesterday. Defeating him is what matters most."

→ ZenX measures your yesterday so you can beat it tomorrow

FEATURED ATHLETE

Vietnam's best.
Using ZenX.

When your platform's most elite user competes at Western States and UTMB, it signals that the intelligence system works at the highest level of the sport.

🏔
Hà Thị Hậu
VIETNAM'S #1 FEMALE ULTRA RUNNER · SUUNTO AMBASSADOR

"Ngày mới bắt đầu, tôi chẳng có gì trong tay. Không đồng hồ, cũng chẳng có đồ chạy xịn. Tôi chỉ biết cắm đầu chạy, chạy và chạy mỗi ngày."

— Hà Thị Hậu · via Suunto Vietnam

From the mountains of Sapa to the world stage — Hậu has proven that Vietnamese athletes can compete and win at the highest level of ultra trail running globally. Her data-driven approach to training and race execution aligns perfectly with ZenX's intelligence philosophy.

🌍 Western States · California 🏔 UTMB World Series 🇻🇳 Sapa Mountain Running Club hathihau.com →
6th
FEMALE
WESTERN STATES
4th
FEMALE
UTMB CCC
1ST FEMALE
RECENT RACES
VERIFIED RACE RÉSUMÉ · 2024–2025
1ST FEMALE
Anta Guanjun HK 100km
Hong Kong · Jan 2026
1ST FEMALE
ChiangMai UTMB Elephant 100km
Thailand · Dec 2025
1ST FEMALE · 11TH OVERALL
Tsaigu Trail 105km
Zhejiang, China · Nov 2025
6TH FEMALE · 18TH OVERALL
Western States 160km
California, USA · Jun 2025
4TH FEMALE
UTMB CCC 100km
Italy · Aug 2024
1ST FEMALE · 3RD OVERALL
Ultra Trail Singapore 50km
Singapore · Jun 2024
2ND FEMALE
Asia Pacific Trail Championship 50mi
South Korea · Oct 2024

EARLY TRACTION

Real athletes.
Real data.

Beta launched within the Vietnamese trail running community — one of the most active in SEA. Early adopters from Chứa Chan Trail Community, Trail Station, and regional race groups.

100+
BETA ATHLETES
340+
TRAINING WEEKS LOGGED
47
RACES ANALYZED (RSI)
3
TRAIL COMMUNITIES
82%
WEEK-2 RETENTION
$0
PAID ACQUISITION COST
COMMUNITIES:
🏔 Chứa Chan Trail Community
⛰ Trail Station
🌿 Vietnam Trail Runners
🏃 CEO Runner Club
+ more joining
AGE RANGE:
20s · Beginners 30s · Intermediate 40s · Elite 54+ · 100km Finishers
ZenX adapts to every athlete profile.
WHAT ATHLETES SAY
FOUNDER
"

Ở ultra càng dài, người đi xa nhất không phải người nghĩ nhiều nhất. Mà là người giữ đầu óc đơn giản nhất. ZenX là hệ thống giúp tôi làm điều đó.

🏔
Hà Thanh Phong
Founder · 100km 1st Age Group · 250km target 2027
"

DNF ở 75km — nhưng đây là race mình hài lòng nhất. Vì lần đầu tiên mình thật sự kiểm soát được quyết định: biết dừng đúng lúc thay vì cố và trả giá. ZenX cho tôi ngôn ngữ để hiểu điều đó bằng data.

DaLat Ultra Trail 100km Pro · CP8 75km · Ankle injury management
Trần Hồng Quân
Ultra runner · Trail coach · 500km Thailand 2025
ELITE ATHLETE
"

Ngày mới bắt đầu, tôi chẳng có gì trong tay. Không đồng hồ, cũng chẳng có đồ chạy xịn. Tôi chỉ biết cắm đầu chạy, chạy và chạy mỗi ngày. ZenX cho tôi thứ tôi không có lúc đó — một hệ thống hiểu cơ thể mình.

Western States 160km · 6th Female UTMB CCC · 4th Female Suunto Vietnam Ambassador
🏔
Hà Thị Hậu
Vietnam's #1 Female Ultra Runner · hathihau.com
"

Trong race thường nói "lần sau đừng đứa nào rủ tao đi 100km nữa"... nhưng khi nghe giá SEB sale 50% là lại mua để dành nửa năm tính tiếp. ZenX giúp tôi hiểu tại sao cơ thể phản ứng như vậy — để lần sau chuẩn bị đúng thay vì đoán mò.

Dalat Ultra Trail 100km · 20h trong rừng · 12,000 calo · Suunto Vertical
🏃
Phạm Ngọc Đức
Co-founder CEO Runner Club · Đồng hành ZenX
"

Ultra trail đêm tối — một phiên Tỉnh Thức để đối diện với chính mình. Điều gì khiến tôi sợ nhất trong đời? Không phải thất bại. Tôi sợ đánh mất sự tỉnh táo khi thành công đến sớm. Giữ mình, để đi xa hơn. ZenX là hệ thống giúp tôi giữ mình đúng cách.

54 tuổi · Dalat Ultra Trail 100km Finisher LAAN Ultra Trail 100km Prenn Trail Summit
🌙
Nguyễn Thanh Đạm
54 tuổi · CCTC member · Multi 100km finisher
💬 More stories from the community · Request access →
GET STARTED

Perform consistently.
Under any condition.

ZenX Ultra is in open beta.
Free to use. Built for serious athletes.
Intelligence grows with your data.

OPEN ZENX ULTRA → EXPLORE ENGINES
Investor or partnership inquiries → hp@zenx.asia