What the AI found
“The AI observed: 'Client (A.S.) shows a significant sleep latency correlation with late-day caffeine intake, specifically an average 37-minute increase in time to fall asleep when caffeine is consumed after 3 PM.'”
Before
Ad-hoc client notes, vague advice
After
Personalised sleep protocols by appointment
The same system, three states — real screens, not a screenshot
- 27 Oct: Slept 11pm-7am. Tea at 4pm. Felt tired.
- 28 Oct: Slept 10:30pm-6:30am. Coffee at 2pm. Good energy.
- 29 Oct: Slept 1am-8am. Espresso at 5pm. Restless.
- 30 Oct: Slept 10pm-6am. No caffeine. Woke refreshed.
Prompt
Here is client A.S.'s sleep diary and caffeine intake for the last month. Please identify any significant correlations between time of caffeine consumption and sleep onset latency (time to fall asleep). Focus on a specific numeric impact if possible, noting any patterns you find: '''27 Oct: Slept 11pm-7am, tea at 4pm. 28 Oct: Slept 10:30pm-6:30am, coffee at 2pm. 29 Oct: Slept 1am-8am, espresso at 5pm. 30 Oct: Slept 10pm-6am, no caffeine. Average sleep latency is 20 min. When caffeine is consumed after 3 PM: 27 Oct: latency 45 min, 29 Oct: latency 55 min. When caffeine is consumed before 3 PM: 28 Oct: latency 25 min. No caffeine: 30 Oct: latency 15 min.'''
Here is client A.S.'s sleep diary and caffeine intake for the last month. Please identify any significant correlations between time of caffeine consumption and sleep onset latency (time to fall asleep). Focus on a specific numeric impact if possible, noting any patterns you find: '''27 Oct: Slept 11pm-7am, tea at 4pm. 28 Oct: Slept 10:30pm-6:30am, coffee at 2pm. 29 Oct: Slept 1am-8am, espresso at 5pm. 30 Oct: Slept 10pm-6am, no caffeine. Average sleep latency is 20 min. When caffeine is consumed after 3 PM: 27 Oct: latency 45 min, 29 Oct: latency 55 min. When caffeine is consumed before 3 PM: 28 Oct: latency 25 min. No caffeine: 30 Oct: latency 15 min.'''
AI
Client (A.S.) shows a significant sleep latency correlation with late-day caffeine intake, specifically an average 37-minute increase in time to fall asleep when caffeine is consumed after 3 PM. Early day caffeine (before 3 PM) shows a smaller, 5-minute increase compared to no caffeine days.| Session Date | 12 Nov |
| Key Observation | Caffeine after 3 PM |
| Latency Impact | +37 min |
From Scattered Notes to Targeted Sleep Protocol
A nutritionist moves from inconsistent client sleep data to actionable, personalised insights for every appointment, using AI to surface hidden patterns.
A nutritionist running a small EU practice, focused on client sleep.
Tools used
The real tools used here — swap any for your own equivalent. Each links to how we’d set it up.
Starting state
Before anything was set up
Before implementing the AI-powered system, client sleep observations were scattered across various notes and manual logs. A.S., for instance, recorded sleep start and end times, general diet, and exercise in a free-form document. This made it challenging to discern specific patterns or quantify the impact of particular behaviours on sleep quality, leading to generalised advice rather than targeted interventions.
- 27 Oct: Slept 11pm-7am. Tea at 4pm. Felt tired.
- 28 Oct: Slept 10:30pm-6:30am. Coffee at 2pm. Good energy.
- 29 Oct: Slept 1am-8am. Espresso at 5pm. Restless.
- 30 Oct: Slept 10pm-6am. No caffeine. Woke refreshed.
Working state
Membership, doing its job
The nutritionist began centralising A.S.'s data into a Google Sheet. By integrating a simple AI prompt within Gemini, she could rapidly analyse weeks of anecdotal notes against structured sleep metrics. This allowed for on-the-fly hypothesis testing and identification of previously unrecognised correlations, transforming raw data into salient client insights that could be integrated directly into their protocol.
Prompt
Here is client A.S.'s sleep diary and caffeine intake for the last month. Please identify any significant correlations between time of caffeine consumption and sleep onset latency (time to fall asleep). Focus on a specific numeric impact if possible, noting any patterns you find: '''27 Oct: Slept 11pm-7am, tea at 4pm. 28 Oct: Slept 10:30pm-6:30am, coffee at 2pm. 29 Oct: Slept 1am-8am, espresso at 5pm. 30 Oct: Slept 10pm-6am, no caffeine. Average sleep latency is 20 min. When caffeine is consumed after 3 PM: 27 Oct: latency 45 min, 29 Oct: latency 55 min. When caffeine is consumed before 3 PM: 28 Oct: latency 25 min. No caffeine: 30 Oct: latency 15 min.'''
Here is client A.S.'s sleep diary and caffeine intake for the last month. Please identify any significant correlations between time of caffeine consumption and sleep onset latency (time to fall asleep). Focus on a specific numeric impact if possible, noting any patterns you find: '''27 Oct: Slept 11pm-7am, tea at 4pm. 28 Oct: Slept 10:30pm-6:30am, coffee at 2pm. 29 Oct: Slept 1am-8am, espresso at 5pm. 30 Oct: Slept 10pm-6am, no caffeine. Average sleep latency is 20 min. When caffeine is consumed after 3 PM: 27 Oct: latency 45 min, 29 Oct: latency 55 min. When caffeine is consumed before 3 PM: 28 Oct: latency 25 min. No caffeine: 30 Oct: latency 15 min.'''
AI
Client (A.S.) shows a significant sleep latency correlation with late-day caffeine intake, specifically an average 37-minute increase in time to fall asleep when caffeine is consumed after 3 PM. Early day caffeine (before 3 PM) shows a smaller, 5-minute increase compared to no caffeine days.Use case implemented
The finished system, running on its own
Now, before each client meeting, the nutritionist runs a quick query on their updated sleep data. The AI provides a concise summary of key behavioural correlations, highlighting actionable areas for improvement. This system ensures every piece of advice is data-backed and tailored, dramatically improving client engagement and the perceived value of the nutritional guidance.
| Session Date | 12 Nov |
| Key Observation | Caffeine after 3 PM |
| Latency Impact | +37 min |
| Recommendation | Shift all caffeine to before 1 PM |
| Next Review | 26 Nov |
What an outside observer would notice
20 minutes
Time saved per client review
100%
Clients receiving personalised sleep protocols
Up 2x
Insights per client session
The stack — build it yourself
Familiar, easily shareable, and adaptable for various client data points.
Provides quick, specific insights from raw text data, enabling fast pattern recognition.
Organises client information and enables development of structured protocols.
These are the tools used in this story. Any can be swapped for an equivalent you already trust.
Go deeper
Do this yourself
See Membership
This story runs on Membership. The tools and prompts above are the real build — swap any tool for your own equivalent and follow the same steps.