How it works

You talk to your AI app the way you would message a coach. Prova Endurance gives it your training history, your thresholds and a library of training research, and checks every plan it writes against training rules.

Who does what

You

You say what you want and how training is going, and you have the last word on your plan.

Your AI app

Claude, ChatGPT or another app that connects over MCP. It talks with you and does the thinking: it writes your plan, explains each session and proposes changes.

Prova Endurance

Keeps your plan, your history and your thresholds. It reads every activity, checks every plan and change against training rules, and holds the research the AI cites. It runs no AI model of its own.

Your platform and device

Your activities come in from intervals.icu or a Garmin Connect export. Planned sessions go out to your intervals.icu calendar, or to your watch or bike computer as a .fit file.

Step by step

  1. 1

    You tell it about yourself

    For example, you say

    I want to finish my first gran fondo in March. I can train about eight hours a week, never on Mondays.

    Your AI starts with the interview a coach would do: your goal and any other races, the hours and days you can train, injuries, the sports you do, and what you know about your own heart rate. Then it reads your training history. With no history, it starts carefully.

    Everything is measured against your own thresholds. The values you give, or accept when Prova Endurance proposes one, are the ones it uses. If you don't know your heart-rate numbers, it starts from the highest heart rate in your data, or from an estimate based on your age and sex, and labels it an estimate. Power and pace thresholds are never estimated: they stay empty until you give one or do a test.

  2. 2

    It writes a plan

    For example, you say

    Build me a plan for March.

    Your AI writes the plan with a reason for every session: the research it rests on, or a plain note that it is the AI's own judgement.

    Prova Endurance checks the plan against training rules that read your weekly hours, how many hard sessions you do and how far apart, and your time at each intensity. If a rule doesn't fit you, you can set your own exception, in your own words. Then the plan waits for you to accept or decline it.

    If a threshold is still missing, a plan of several weeks names a test that would measure it. After you do the test, Prova Endurance reads it and proposes the value.

  3. 3

    You train

    For example, you say

    Send me Thursday's intervals for my bike computer.

    Accepted sessions go to your intervals.icu calendar, or to your device as a .fit file, each with what it is for and the workout step by step.

    Targets come from your thresholds, or from your own zones if you give your AI your zone table. Prova Endurance has no zone table of its own.

    Once a plan is running, your AI can prescribe each next week in a scheduled session. That week is applied without waiting for you and emailed to you with its reasons, and you change it afterwards by asking.

  4. 4

    It reads what you did

    For example, you say

    Heavy legs today, and it was hot. I'd call it a 7 out of 10.

    Each activity comes in from your platform and is matched to the session you planned, so you see what you did next to what was planned. Tell your AI how it felt and it keeps your words and your rating with the activity.

    Prova Endurance describes each session against your thresholds: the time you spent below your first threshold, between the two, and at or above your second, read from power or pace and from heart rate, side by side. Each week is compared with your own recent weeks and read as building, steady, backing off or spiking, with the numbers behind the word.

  5. 5

    It adapts

    For example, you say

    I was sick and lost four days. The race is still on.

    Tell your AI what changed and it proposes a new version of the plan, checked against the same rules, for you to accept or decline.

    Your resting heart rate and HRV are read as an average of several days against your own baseline. When they drift far from it, your AI brings it up with you, and the reading alone never changes your plan.

What the advice rests on

A training-load score multiplies how long you trained by how hard, which hides which of the two it was. A review of the concept found that training load has never been validated as a measure of training dose, and suggests it may be better to show the time spent in each intensity domain separately. That is how Prova Endurance shows every session. [1]

Your two thresholds split each sport into three intensity domains: below the first, between the two, and at or above the second. A study of how elite endurance athletes train split their training the same way. [2]

The rules count hard sessions because a review of endurance training found about two a week enough, with more tending toward overreaching. [3]

An estimated maximum heart rate comes from an age-and-sex formula whose authors say its error has to be taken into account. That is why it is labelled an estimate, and why a maximum you give, or the highest heart rate in your data, takes its place. [4]

Resting heart rate and HRV are read as averages because a review reports that, in endurance athletes, they tracked running performance only when averaged over at least three to four days, not as measures taken on single days. [5]

What you say comes first. A systematic review found that athletes' own reports of how they feel reflected their training more sensitively and consistently than common objective measures. There is no readiness score. [6]

Your fitness is your thresholds over time. Settings shows each value, when it took effect and where it came from. There is no fitness or fatigue score.

Research cited

  1. [1] Passfield L, Murias JM, Sacchetti M, Nicolò A. Validity of the Training-Load Concept. Int J Sports Physiol Perform. 2022;17(4):507-514. doi:10.1123/ijspp.2021-0536 DOI
  2. [2] Seiler KS, Kjerland GØ. Quantifying training intensity distribution in elite endurance athletes: is there evidence for an 'optimal' distribution? Scand J Med Sci Sports. 2006;16(1):49-56. doi:10.1111/j.1600-0838.2004.00418.x DOI
  3. [3] Seiler S. What is best practice for training intensity and duration distribution in endurance athletes? Int J Sports Physiol Perform. 2010;5(3):276-291. doi:10.1123/ijspp.5.3.276 DOI
  4. [4] Nes BM, Janszky I, Wisløff U, Støylen A, Karlsen T. Age-predicted maximal heart rate in healthy subjects: the HUNT Fitness Study. Scand J Med Sci Sports. 2013;23(6):697-704. doi:10.1111/j.1600-0838.2012.01445.x DOI
  5. [5] Buchheit M. Monitoring training status with HR measures: do all roads lead to Rome? Front Physiol. 2014;5:73. doi:10.3389/fphys.2014.00073 DOI
  6. [6] Saw AE, Main LC, Gastin PB. Monitoring the athlete training response: subjective self-reported measures trump commonly used objective measures: a systematic review. Br J Sports Med. 2016;50(5):281-291. doi:10.1136/bjsports-2015-094758 DOI

Start today. It is free.

An AI app that can connect to Prova Endurance, and a Google account.

Get started