Coach Chat Privacy in N+One: What the Model Sees and What It Forgets
5 min read
Learn what Coach Chat sees during a session, what it does not remember by itself, and how to redact, share, delete, or save training context on purpose.
On this page
Photo by Ben Guernsey on Unsplash.
Coach Chat processes what you send in session, but it does not keep long-term memory unless account features store history.
This explainer covers what the model can see during a Coach Chat session, what it does not remember by itself, and how you can share useful training context without handing over more detail than needed.
Photo by Buddy AN on Unsplash.
During a live Coach Chat session, the model works from the text, files, and metrics you send. If the app adds recent ride context or profile fields, those inputs can also shape the reply.
Think of the chat as a workbench, not a sealed diary. The answer you get depends on the context placed on that bench during the session.
For better answers, ask focused questions and give the smallest useful training sample. The guide to better prompts for cycling answers can help you frame that context without pasting a whole data dump.
If you want to understand why a session was suggested, compare the chat output with why-this-session notes. Those notes help separate training logic from the private details you choose to share.
Paste only the workout facts needed for the question.
Remove names, account IDs, and private labels first.
Use summaries before raw exports when summaries are enough.
Check attached files before you send them.
This keeps the session useful while limiting what the model sees.
In N+One terms: the coach reads your notebook while you are talking; it does not peek at locked files unless you attach them.
A chat model does not, by itself, carry a private memory of you across every future session. Long-term recall depends on product features that save history, notes, or profile context.
That distinction matters. The model processing your message is not the same thing as your account storing a transcript or saved coaching note.
When you review coach memory across months, look for which facts are saved by the product layer. Stored context can help coaching, but it should be clear and under your control.
If an app shows past chats, saved notes, or persistent preferences, treat those as records. Use the same care you would use with a training log that contains personal details.
Do not assume every chat is saved forever.
Do check whether history is turned on.
Delete chats you do not need to keep.
Keep a local note of advice if you need a record.
The clear next move is to check what is saved, then keep only what helps.
In N+One terms: forgetting is the default for the model; saving is a product action you should be able to see and manage.
Data analysis is free, forever. The Dynamic Coach is Pro. Connected to Garmin, Strava, Whoop, and intervals.icu.
One tactical email with training ideas and product updates. No spam — unsubscribe anytime.
What the model sees: everything you type into a live Coach Chat session during that session.
Photo by Olena Bohovyk on Unsplash.
Share the training signal, not the full identity trail. Power, heart rate, duration, perceived effort, and weekly load can often answer the coaching question without names or account fields.
Do not paste passwords, government IDs, insurance details, or full medical records into Coach Chat. If health context matters, use a short anonymized summary and keep it limited to the question.
For app basics, pair this rule with how to use AI Coach chat and profile and goal settings. Your profile can hold stable training context, while chat can handle the current question.
When a recommendation feels too broad, ask a sharper follow-up instead of adding private detail. The goal is enough context for a good coaching choice, not every fact available.
Share weekly summaries before raw files.
Use age ranges instead of exact birth dates.
Remove rider names from exports when possible.
Do not paste full medical records.
Use synthetic examples for sensitive cases.
In N+One terms: give the coach the training signals it needs, not the keys to your account.
Keep reading
- Sync Latency in N+One: How Long After You Hit Stop Until Your Coach Sees It — No peer-reviewed source gives an exact N+One sync latency number. Learn what can delay coach visibility and the one manual resync test to run now.
- The Personalisation Layer in an AI Cycling Coach: How It Learns Your Quirks — Learn how the personalisation layer in an AI cycling coach adapts generic training logic to your rides, recovery signals, preferences, and device qui...
- Easy AI Cycling Coach — How N+One Makes Coaching Accessible — N+One is an easy AI cycling coach that turns complex training science into simple, personalized daily guidance. Learn how conversational AI, adaptive...
Start with the settings you can see. Look for conversation history, transcript export, clear history, delete, and account privacy settings.
If a retention window is not stated, ask support for the exact policy in plain words. You need to know what is stored, how long it stays, and how deletion works.
If you are new to the product, app navigation basics can help you find the right area faster. For a broader view of the system, see how N+One’s AI coach works.
Set one default rule after checking. For example, keep history off unless you need a saved advice trail for a short block.
Open Coach Chat settings today.
Check whether history is enabled.
Export advice you want to keep.
Delete chats that are no longer useful.
Ask support if storage terms are unclear.
Day 0 — Quick settings check: Open Coach Chat settings. Note whether conversation history or transcripts are enabled and the stated retention period. Turn off history if you prefer not to save sessions.
Day 1 — Redact and share: Prepare one coaching prompt with a weekly ride summary or key metrics. Remove identifying fields before you paste it into chat.
Day 3 — Confirm storage: Check the conversation list. If the chat was saved and you do not want it stored, use delete or clear history where available.
Day 5 — Export and archive: If you want a local copy, save only the coaching advice in your own log. Avoid keeping raw transcripts with personal details.
Day 7 — Set your rule: Pick one default rule for future chats, such as always redact, never save history, or review saved chats on a set schedule.
Coach Chat processes what you send in session, and storage depends on account features you can check. Your best move is simple: redact first, share only the training context needed, then delete or save the transcript on purpose.
No. Treat the live chat as seeing what you type, paste, attach, or what the app explicitly supplies as context. If you are unsure what context is included, check settings or ask support.
The model does not inherently keep a private memory of you between sessions. Any ongoing recall depends on product features such as saved history, notes, or profile context.
Remove names, account IDs, exact birth dates, private labels, and unrelated personal fields. Keep the training facts needed for the coaching question.
Yes. For many coaching questions, a clear summary of ride duration, effort, power, heart rate, and recent workload is more useful than a raw file full of identifiers.