Your Endurance Journey Buddy
Ordinary people can achieve extraordinary goals.
TriBuddy is an AI-powered endurance companion. This anonymous product tour shows how training data, athlete context, a virtual coach, weekly planning and professional evidence can support one deliberate journey—without sharing a private athlete account.
Illustrative data shaped from real TriBuddy workflows · no personal account data, live AI calls or private training records are accessed.Why TriThree disciplines. Three qualities.Swimming · Cycling · RunningCourage · Commitment · Transformation
Every journey needs a buddy
TriBuddy began with swimming, cycling and running, but the journey is really about courage, commitment and transformation.
The example below follows one private 70.3 preparation cycle from training evidence to the next useful action.
01 · Training data
Record the work. Keep its evidence visible.
Imported history, FIT telemetry and athlete feedback form one clear decision context.
42-day rhythm
Volume with recovery space
EarlierToday
Activity detail
Low-cadence tempo intervals
Cycling · 68.4 km · 2h 20m · 29.6°C
Training intentPrimary intent · Power · 224–252 W (80–90% FTP)
Athlete feedbackSession RPE 7 · muscular RPE 8 · quality 4 / 5
Targets and perceived response establish the baseline for the review below.DeepSeek
Virtual Coach · weekly review
Build the next week around repeatable aerobic quality.
The example week is balanced, but the highest-value decision is protecting the recovery window after the key ride so the long run stays aerobic rather than becoming accumulated fatigue.
- Load is concentrated in two quality sessions
- Readiness is adequate, not exceptional
- Heat exposure should change execution, not the purpose
02 · Training calendar
See completed work and planned intent in one place.
A month view keeps imported activity and future training distinguishable at a glance.
August 2026 Activity Planned session
MonTueWedThuFriSatSun
Completed sessions can link to their detail view; plans remain draft, confirmed, completed or skipped—never silently changed by AI.
03 · Activity detail
Review the session against its intended purpose.
Aug 4 · Cycling
Low-cadence tempo intervals
Cycling · 68.4 km · 2h 20m · 29.6°C
Time series · session output
Power held within the intended tempo band.
The shaded band is the 80–90% FTP target; cadence is intentionally lower during the three work blocks. Power Cadence 80–90% FTP target
Time series · physiological response
Heart rate rises without a sharp late-session break.
Compare this response with power, cadence, temperature and perceived muscular load—not in isolation. Heart rate CadenceRead changing cost against a stable workload.
Intervals imported
Targets beside outcome
Warm-up20:00
Easy · < 157 W 146 W · 89 rpm
Tempo 115:00
224–252 W · 60–70 rpm 236 W · 66 rpm
Tempo 215:00
224–252 W · 60–70 rpm 232 W · 64 rpm
Tempo 315:00
224–252 W · 60–70 rpm 228 W · 63 rpm
Cool-down18:00
Easy recovery 139 W · 91 rpm
Sustained output
Power-duration profile
5 sec576 W
1 min318 W
5 min254 W
20 min228 W
Intensity distribution
Time in cycling zones
Z122%
Z241%
Z325%
Z4+12%
Efficiency and heat
Interpret the session in context
- Power / HR drift
- 3.8% Below 5% watchline
- Device temperature
- 29.6°C Heat-aware execution
- Cadence trend
- −3 rpm Across work blocks
DeepSeek
Virtual Coach · activity review
All three work blocks reached target power, while cadence fell slightly across the set.
Interpretation
The small efficiency drift is acceptable for a warm tempo session. The muscular RPE of 8 makes low-cadence torque—not aerobic capacity—the limiting signal to watch.
Next useful step
Manual request only · framed around the plan, telemetry, athlete feedback and recent training context.Keep the next ride easy. Repeat this format only when the legs feel fresh; progress duration before raising target power.
04 · Athlete profile
Give every recommendation an athlete-specific baseline.
Availability, goals, thresholds and sport-specific zones constrain the planning and review context.
Training reference
Preferences that shape the week
Race focus70.3 Triathlon
Weekly availabilityTue · Thu · Fri · Sat · Sun
Long sessionSunday
Training priorityBalanced
Threshold history
Active values by sport
Cycling FTP280 WPower zones active
Run threshold4:30 /km · 165 bpmPace + HR zones active
Swim CSS2:10 /100 mCSS pacing active
Active zones
Clear targets, not generic advice
Run aerobic140–147 bpm
Bike endurance157–210 W
Swim aerobic2:18–2:26 /100 m
05 · AI Virtual Coach
Turn the data into a useful next decision.
The virtual coach is asked manually and evaluates purpose, execution, athlete feedback and recent context. It prioritizes gaps and practical next steps instead of repeating every metric. Its scope is deliberately visible: it explains its coaching logic, identifies uncertainty and leaves every decision with the athlete.
The useful finding
Pace was controlled through the final repeat.
Heart rate rose gradually while pace stayed within the intended threshold range; this supports aerobic durability rather than an all-out effort.What to improve
Protect recovery before the next quality session.
The late-session heart-rate cost suggests keeping the following day easy, especially if heat or poor sleep repeats.Longer-term focus
Build repeatable threshold volume.
Progress one variable at a time: add a repeat or extend the work block only after two controlled sessions at the current demand.Decision boundary Coaching guidance remains reviewable. It does not diagnose injury, silently change training or replace professional care.
06 · AI Planning
Propose a week that remains under athlete control.
Plans use availability, thresholds, training priority and heat-aware execution notes.
DeepSeek
Manual proposal
AI drafts are selectable, editable and never saved automatically.
Partial week sample4 selected sessions shown · Aug 3–Aug 9, 2026
Bike · 75 min · steady aerobic work
AI draftAerobic ride
Warm-up: 15 min at 140–160 W, including 3 × 30 sec at 100–110 rpm with 60 sec easy. Main: 3 × 15 min at 175–195 W and 85–95 rpm, with 3 min below 155 W between blocks. Cool-down: 9 min easy. Start early or move indoors if heat is high.
Primary intentPower · 157–210 W
Secondary intentCadence · 85–95 rpm
Swim · 50 min · about 2,200 m
AI draftTechnique + CSS
Warm-up: 300 m easy, then 4 × 50 m drill with 20 sec rest. Main: 3 rounds of 4 × 100 m at 2:18–2:26 /100 m with 20 sec rest, followed by 100 m very easy. Cool-down: 200 m relaxed. Keep the stroke long and breathing controlled; stop chasing pace if form shortens.
Primary intentPace · 2:18–2:26 /100 m
Secondary intentSkill · relaxed breathing
Run · 55 min · controlled progression
AI draftProgression aerobic run
Warm-up: 12 min at 5:45–5:55 /km. Main: 20 min at 5:25–5:35 /km, then 10 min at 5:10–5:20 /km. Finish with 6 × 20 sec relaxed strides at 4:05–4:25 /km and 60 sec easy jog recoveries. Cool-down: 5 min easy; keep the progression controlled rather than sprinting.
Primary intentPace · 5:45–5:15 /km
Secondary intentHeart rate · < 150 bpm
Run · 80 min · cool-hour preferred
AI draftLong aerobic run
Warm-up: 15 min at 5:50–6:05 /km. Main: 50 min at 5:30–5:45 /km and 140–147 bpm. If heart rate remains stable, add 10 min at 5:20–5:30 /km; otherwise stay aerobic. Cool-down: 5 min easy. Take 30–45 g carbohydrate per hour and reduce pace before extending duration in the heat.
Primary intentPace · 5:30–5:45 /km
Secondary intentHeart rate · 140–147 bpm
DeepSeek
Virtual Coach signal used by Planner
Why this draft protects the next useful adaptation.
The recent low-cadence session was muscularly demanding, so the draft delays another torque-focused ride and keeps the long run aerobic. The quality stimulus is retained through swim CSS and controlled run progression rather than adding a second hard bike day.
Referenced coach signal · muscular RPE 8 after low-cadence intervals07 · Extended Study
Turn a coaching question into a learning loop.
Extended Study is intentionally separate from coaching automation. It makes the research path inspectable: define the question, search multiple scholarly sources, preserve provider metadata, validate identity and then read or save evidence with its limitations still attached.
Manual retrieval logic
One question becomes a source-aware evidence set.
Every available branch searches independently. A service issue stays visible and never masquerades as zero candidates.
DeepSeek
Frame one focused questionDeepSeek prepares an editable structured query only when manually requested.
Merge and rankDeduplicate DOI and title matches · retain provider identity · combine source rankings · give Scopus the strongest individual weight when available
- 1Inspect candidates
- 2Select and enrich
- 3Save or mark as Key Read
Suggested question
How does low cadence during cycling intervals affect power sustainability and the quality of the subsequent run in triathletes?
Origin: a manually selected training pattern or coach-suggested learning topicSearch professional sources
TITLE-ABS-KEY((triathlon OR cycling) AND cadence AND interval* AND (power OR fatigue) AND (run OR performance))The editable query runs across every available source. Scopus supplies the preferred curated signal when available; OpenAlex, CrossRef and Europe PMC keep discovery operational in the free-service workflow.Merge and enrich evidenceCadence selection and physiological response during cycling exerciseDeduplicate identifiers, preserve every discovery source and independently check permitted metadata, abstracts and lawful-access locations.
Abstract · enriched evidence
The evidence card preserves the provider-recorded abstract and source metadata. The athlete-facing interpretation is deliberately separated from the source text, so an individual session decision remains provisional and context-dependent.
Open lawful source ↗Save the useful evidence
Place exceptional papers in Key Reads; group the rest of the Saved Library by research question, search it and track reading status without relying on subjective star rankings.
After discovery
Turn a merged result set into evidence you can trust.
The retrieval tree explains where candidates come from. This next layer preserves identity, exposes gaps and keeps scientific evidence separate from coaching judgment.
Identity preserved
Know which source supplied each fact.
Provider, DOI, title, authors, journal and publication year remain attached to the candidate after merging, so one combined result never becomes an opaque record.
Availability disclosed
A service issue never becomes zero evidence.
Each source reports its own candidate count and retrieval status. An unavailable provider stays visibly unavailable instead of silently weakening the result set.
Evidence boundaries
Missing source text is never invented.
Abstracts and lawful-access locations are shown only when a permitted source returns them. Translation is reversible and remains separate from the original text.
Human judgment
Research informs; it does not prescribe.
A paper can refine a coaching question or future planning signal, but it never changes an athlete plan automatically and is always interpreted against the athlete's context.
From evidence to a durable reading library
Promote exceptional papers; organize everything else.
Key Read is a deliberate prominence decision, not a star rating. The Saved Library keeps supporting evidence searchable, grouped and ready for later comparison.
Key Read
PinnedCadence selection and physiological response during cycling exercise
Pinned example · directly relevant to the active questionA Key Read is promoted because it materially improves understanding of the question—not because it received a subjective star score. The source identity, accessible abstract and lawful reading location remain attached.
- Pinned above the ordinary library
- Original and translated abstract
- Question and source provenance retained
Saved Library
OrganizedTraining intensity distribution in endurance athletes
Useful context · retained for later comparisonUseful evidence remains grouped by research question even when it is not a defining paper. Search and reading status keep a growing library manageable without forcing every record into the Key Read tier.
- Grouped by research question
- Searchable title, author and journal
- Unread, Reading and Read status