Documentation AI Coaching

AI Coaching

GoAnd.Run uses a large language model to act as your running coach. It reads your training data: activities, zones, splits, race history, and profile, and generates targeted coaching feedback. There are five types of AI output: session feedback on individual activities, weekly reviews after training weeks, training plan generation for your race goals, on-demand session adjustments, and AI-generated strength & conditioning workouts.

Session Feedback

On-demand analysis of a single activity. How did this run go and what should you take from it?

Weekly Reviews

Summary of the training week, progress assessment, and guidance for the coming week.

Training Plans

Full structured training plan built around your race goal and current fitness.

Session Adjustments

On-demand alternative when you cannot do today's planned session. Replaces one day without touching the rest of the plan.

S&C Workouts

Structured strength & conditioning sessions tailored to your training phase, equipment, and injury history.

Session Feedback

Session feedback is generated on demand from the activity detail page. You request it when you want it (it is not automatic), which keeps costs predictable.

What the AI analyses

Activity metrics: distance, duration, average pace, heart rate, elevation, and cadence if available.

Zone distribution: how time was distributed across the five heart rate zones.

Splits: per-kilometre pace and heart rate to assess pacing strategy.

Recent training context: the activities in the days before, to assess cumulative fatigue and training load.

Athlete profile: your age, experience level, and coaching style preference.

What the feedback includes

The session feedback is structured around three areas: what went well, what to watch or improve, and a specific recommendation for your next similar session. Feedback is personalised to your coaching style. A data-driven style will include more numbers and metrics; a supportive style will be more conversational.

Weekly Reviews

Weekly reviews are a summary of a full training week. They are generated automatically when you submit a weekly check-in from your training plan page, giving the AI both the objective data from your activities and the subjective context from your check-in notes.

What the weekly review covers

Training summary

Total distance, time, and number of sessions. How this compares to the previous week and to the planned week.

Training quality assessment

Zone balance across all sessions: was it a high-quality week or was there too much junk miles in Zone 3?

Progress towards goal

Based on current form and training adherence, assessment of whether you are on track for your race goal.

Guidance for next week

Specific adjustments or focus areas for the upcoming training week based on how this week went.

Where to find weekly reviews

Weekly reviews appear on your training plan page, associated with the relevant week. They are also accessible from the Dashboard, which surfaces the most recent review. Reviews are stored permanently. You can look back at any week's review at any time.

S&C Workouts

When strength & conditioning is enabled in your preferences, training plans include dedicated strength days alongside your running sessions. On those days you can generate a fully structured S&C workout from the session detail page, tailored to your training phase, available equipment, and injury history.

Enabling S&C workouts

In your preferences, toggle on "Generate personalised S&C workouts". This tells the plan generator to include strength days as structured sessions rather than rest days.

Select your available equipment: Bodyweight Only, Resistance Bands, Dumbbells & Kettlebells, or Full Gym. The AI constrains exercise selection to what you actually have.

The next time you generate a training plan, strength days will appear in the schedule. Open a strength day session page and click Generate Workout to get your S&C session.

What the workout includes

Warm-Up

Dynamic mobility and activation exercises to prepare the body and reduce injury risk. Typically 5–10 minutes.

Main Set

The primary strength work: exercises with specific sets, reps, and coaching notes. Volume and intensity are phase-appropriate.

Cool-Down & Mobility

Stretches and mobility work to aid recovery and maintain flexibility relevant to running.

Phase-aware guidance

S&C intensity adapts across your training cycle. In the base and build phases, workouts are designed to develop strength: higher loads, compound movements, progressive overload. As you move into the sharpen and taper phases, the volume and intensity reduce so that the S&C work supports rather than compromises your key running sessions. During race week, strength sessions are skipped or limited to light activation only.

Injury awareness

Any injuries or niggles you report in your weekly check-in or training profile are passed to the S&C agent. If you have flagged a sore achilles, for example, exercises that load that area will be excluded or replaced with lower-risk alternatives.

Workouts are generated once per session. Once generated, the workout is stored and displayed instantly on every subsequent visit. You can regenerate a workout for a session if you want a fresh variation; this will replace the previous one.

How the AI Uses Your Data

Understanding what data goes to the AI helps you make the most of the coaching features and gives you confidence in how your information is used.

Data sent with each request

GoAnd.Run assembles a coaching context before sending any request to the AI. This context includes:

For session feedback

The full activity metrics, zone distribution, splits, your athlete profile, and a summary of the preceding 7 days of training.

For weekly reviews

All activities from the review week, your check-in notes, the planned sessions for the week, and a summary of recent weeks for context on training trends.

For training plans

Your race goal, athlete profile, recent 8 weeks of training volume and zone distribution, recent race results, and available training days.

For session adjustments

The planned session, your stated reason, the full week's sessions, activities completed so far this week, your athlete profile and pace zones, and recent session feedback.

For S&C workouts

The strength session details from the plan, the full week's sessions, your training phase, equipment preference, injury history, and the running sessions immediately surrounding the strength day.

What is never sent

Your Strava access token or credentials
Your email address or account password
GPS route data or map coordinates
Raw activity streams (full second-by-second data)

AI Usage & Costs

AI features consume tokens from the underlying language model. GoAnd.Run tracks your token usage and estimated cost. Navigate to AI Usage from your account menu to see your usage history.

Understanding the usage dashboard

Input tokens

Tokens used by the context sent to the AI (your training data, profile, and the coaching prompt).

Output tokens

Tokens used by the AI's response (the feedback or plan text itself).

Cost per request

The estimated cost in USD based on the model's pricing. Session feedback is typically 0.5–2 cents. Training plans are typically 3–10 cents due to the larger output.

Running total

Total cumulative cost since you started using GoAnd.Run. Stored responses do not incur additional cost to view.

Feedback is cached. Once feedback or a weekly review is generated and stored, viewing it again costs nothing. The cost is incurred only at the time of generation.

Limitations

AI coaching in GoAnd.Run is a powerful tool, but it has important limitations you should understand before relying on it for serious training decisions.

Not a substitute for medical advice

GoAnd.Run cannot diagnose injury, illness, or overtraining syndrome. If you have physical symptoms, consult a sports medicine professional or physiotherapist before continuing to train.

Data quality affects feedback quality

The AI is only as good as the data it receives. Activities without heart rate data produce less useful analysis. Incorrectly configured HR zones will lead to misclassified zones and potentially misleading feedback.

AI can be wrong

Large language models can produce plausible-sounding advice that is incorrect or not suited to your individual situation. Use the feedback as one input into your training decisions, not as a definitive prescription.

Plans are a starting point

Training plans generated by GoAnd.Run are structured suggestions. Listen to your body. If a session feels wrong or you are carrying accumulated fatigue, it is always acceptable to modify or skip a session.