Decis
Log InStart Free
Decis80+ datasets · MCP · CLI · API

If you stop coffee before 2 PM, will your sleep change?

Decis normalizes years of records into standard datasets, then finds and compares the days that resemble the habit you want to change. Ask in Decis, or connect Claude or ChatGPT and analyze them your way.

Download on the App Store
QR code to install the Decis iOS appScan with your iPhone camera to install the app.

Decis for self-trackers

In the sample records, the four weeks after the change looked like this.

Once you have tried it, Decis compares before and after and tells you which records it checked.

Actual product UI · de-identified sample records
More deep sleep after an earlier caffeine cutoff
+9m
Less time awake during sleep over the same period
−8m
Caffeine logs over 56 days behind this comparison
143 logs

These differences come from de-identified sample records, and they do not show a cause. Connect your records to see your own numbers.

01Ask

Answers from your records

Ask for a before-and-after, and the answer says which records it checked.

Each answer runs from conclusion to evidence, interpretation, next action, and data coverage. It names the missing days and the other things that changed at the same time, so you can judge how far to trust your own experiment.

  • Five steps from conclusion to data coverage
  • Correlation and sample limits stated in the answer
  • Answers that cannot reach a conclusion do not use your allowance
AskActual product UI · de-identified sample records
Ask Decis5 answers left this month5 leftAnalyzable daily records: Jan 1, 2025–Jul 17, 2026Analyzable daily records2025-01-01–2026-07-17
You

I moved my caffeine cutoff to 2 PM on June 1. Compare my sleep in the four weeks before and after

Decis

Conclusion

In the four weeks after you moved your caffeine cutoff, your records show less time awake during the night and slightly more deep sleep. Total sleep barely changed, so the difference is in how the sleep was made up, not how long it was.

Evidence

  • Your last logged caffeine moved from 4:40 PM to 1:25 PM on average, and daily intake fell from 248mg to 221mg.
  • Deep sleep rose from 52 to 61 minutes on average, and time awake during sleep fell from 41 to 33 minutes.
  • Total sleep moved from 6h 58m to 7h 03m, a five-minute difference.
  • You missed the cutoff on five days in the later period. Deep sleep on those five days averaged 50 minutes, close to the earlier four weeks.

Interpretation

The split between days you kept the cutoff and days you missed it, inside the same period, suggests the change is unlikely to come from season or schedule alone. It is still a before-and-after comparison, which shows association. You also went from three workouts a week to four in June, and these records cannot separate the two changes.

Next action

Consider keeping the cutoff for two more weeks while holding your workout count steady. Moving the cutoff back to 4 PM for one week after that would show whether the same difference appears again.

Data coverage

Reviewed 56 days from May 4 through June 28, including sleep stages, 143 caffeine logs, and 27 workouts. Caffeine logs were missing on three days in the earlier period and one day in the later period.

Interpreted from your records · correlation may not mean causation
02Timeline

24-hour timeline

See the whole day in sequence.

See heart rate, HRV, sleep, training, events, and meals on one 24-hour axis. Open a meal to see its macro balance, key nutrients, and your running daily total at that point.

  • Heart rate, HRV, sleep, training, events, and meal lanes
  • Before-and-after context on one shared time axis
  • Open meals for macro balance, key nutrients, and daily progress
Loading the actual Timeline view
03Today

Today's records

See today’s wellness patterns at a glance.

Compare HRV, sleeping heart rate, and sleep duration with your usual range, then follow sleep stages and the balance between long-term fitness and recent fatigue on the same screen.

  • HRV and sleeping heart rate against your recent range
  • Sleep stages and the latest 14-day sleep trend
  • Fitness, fatigue, form, and daily training load
Actual product UI · de-identified sample records
Condition

Your recovery readings are in your usual range, and you slept well.

Today's conditionRoom to pushAll signals are within your usual range
Fitbit HRV · RMSSD (sleep)
Apple Health HRV · SDNN
HighAvgLowVery low
Sleep HRV
59msUsual 53–59

Based on Fitbit HRV · RMSSD (sleep)

Sleep HR
49bpmUsual 48.5–51.5

The morning recovery signal — resting HR arrives a day later

Resting HR
53bpmUsual 52.5–55.5

Your latest all-day resting heart-rate reading

This observation compares HRV and sleeping heart rate with your usual range over the last 60 days, and sleep with your personal sleep need. It does not determine safety or a medical condition.

04Calendar

Your schedule and the next day

Find which events leave you slow to recover the next day.

For every recurring event, like the weekly operations review, Decis gathers next-day HRV, sleeping heart rate, and sleep and shows how often it went that way, such as 6 of 8 times. You also see which weekdays and hours your meetings cluster in. A pattern is shown as a pattern, never as a cause.

  • How your body was after each recurring event, with the observation count
  • Next-day HRV, sleeping heart rate, and sleep duration
  • Meeting density by weekday and hour
Actual product UI · de-identified sample records
Last 3 months energy vampire · recurring event

Weekly operations review

On 6 of 8 occasions, next-day recovery moved in the same direction.

Apr 1, 2026–Jun 30, 2026
Repeated6 of 8 timesConsistently repeated
HRV-8 ms
Sleeping heart rate+4 bpm
Sleep duration-32 min
Basic stats

Your Last 3 months schedule in numbers

Covers Apr 1, 2026–Jun 30, 2026. Meeting figures include events classified as meetings; location figures include all events.

Meetings46 times
Meeting time49 hr 40 min
People met18 people
Days with meetings31 days
In person 13Video 27Phone 2Other 4
Busy meeting times
9:0012:0015:0018:00
Mon
Tue
Wed
Thu
Fri
Sat
Sun
Your busiest day was May 12, 2026: 5 meetings totaling 5 hr 10 min.

In the AI you already use

The Claude or ChatGPT you already use reads your records and answers.

Paste one address and you are done. Your AI queries more than 80 datasets, from sleep and recovery to meals, calendar, and agent work, and answers from your own numbers for that day. Read-only, and you can disconnect it on the Agents page at any time.

Claude · Decis connected

Can I run intervals tonight? I was reviewing code late last night.

Read from Decis · today’s brief, 14 days of sleep and HRV, yesterday’s agent sessions

Your records point to easing off one notch today.

  • 5h 48m of sleep last night against a 14-day average of 6h 52m
  • Morning HRV 43ms, 16% below your 51ms baseline
  • Yesterday’s Claude Code session ended at 23:40, and on all 5 near-midnight work days in the last 4 weeks, next-day HRV sat below baseline

Swap the intervals for 40 minutes easy and check tomorrow’s HRV.

De-identified sample records · days with the same conditions compared, no cause claimed.
  1. 01
    Paste one addresshttps://www.decis.me/mcp

    Works in Claude, ChatGPT, Cursor, or any app that supports MCP.

  2. 02
    Approve once in your browser

    You see what will be read before allowing it. There is no edit or delete permission to grant.

  3. 03
    Ask in plain words

    “Why is my sleep off this week?” “Should I cut today’s workout?” “How was the day after yesterday’s coding session?”

In the terminal and coding agents, use the CLI
brew install tab0inc/tap/decis-cli
decis data list --domain sleep
decis work connect pi

Connect Claude Code, Codex, and Pi work logs and read the same 80+ datasets straight from the terminal.

  • Claude
  • ChatGPT
  • Claude Code
  • Cursor
  • Decis CLI
  • Any MCP client

  • Read-only by design

    Connected AI apps can read your data but can never modify or delete it.

  • You consent, and you can revoke

    Each app is approved in your browser, and revoking it on the Agents page removes its access immediately.

  • 80+ datasets on a stable contract

    Sleep, nutrition, training, recovery, and schedule data served through a discoverable catalog and an OpenAPI contract.

How to connect and disconnect

Questions self-trackers ask

Review your own experiments against your records.

Selecting a question fills the Ask composer first. Observed differences are correlations, and each answer says when the sample is small or other factors may be mixed in.

Answers that cannot reach a conclusion because records or sample size are insufficient do not use your question allowance.

Connections and privacy

Read only what you allow. Leave the originals untouched.

Connect Apple Health through the iOS app, Google Health by approving access with your Google account, and Google Calendar for read-only schedule context.

  • Apple HealthSync sleep, heart rate, HRV, activity, training, and nutrition through the iOS app
  • Google HealthSleep, heart rate, HRV, activity, and nutrition from supported devices
  • Google CalendarRead-only schedule context for Calendar insights and Ask
D
DecisToday’s wellness patterns from your connected records
Read only

Review requested permissions before connecting. Decis does not change records in connected services, and you can manage source-specific synced data and your Decis account data in Settings.

  • Minimal, read-only access

    Decis requests only the permissions needed for the feature and does not change originals in connected services.

  • Protected in transit and at rest

    Data in transit, stored data, and external-service connection credentials are encrypted.

  • Manage connections and deletion

    Remove a connected source or request permanent deletion of your Decis account in Settings. Applicable retention rules are described in the privacy policy.

See how Decis handles data

Frequently asked questions

What to know before you begin.

The essentials about these product views, connections, answers, privacy, and health and wellness information.

Are these the actual product screens?

Yes. The UI components and visualizations used in Today, Timeline, Calendar, and Ask are rendered here from the same product code. The dates, values, and event names shown are public de-identified examples, not a user’s records.

Which records can I connect?

You can connect sleep, heart rate, HRV, activity, training, and nutrition records from Apple Health through the iOS app or from Google Health, and connect Google Calendar for read-only schedule context.

What happens when there is not enough evidence?

If records or sample size are insufficient for a conclusion, Decis explains what is missing and which records would help. That answer does not use your question allowance.

How is health data protected?

Decis uses read-only access within the scope you approve and encrypts data in transit, stored data, and connection credentials. You can manage connected sources and account data in Settings.

Can Decis give medical advice?

Decis organizes everyday wellness records and presents observed patterns. It does not provide medical diagnosis or treatment. Confirm important health and safety decisions with a qualified medical professional.

Where Decis is heading

How far can your records take you?

Today Decis compares your past records to help with today's choice. The more records you have, the closer they come to a model of you, and the goal is to use that model to judge a change before you make it.

  1. 01Now

    Gather

    Sleep, training, meals, calendar, and AI work records on one timeline.

  2. 02Now

    Compare

    How days like that differed from days that were not, with the sample size and limits shown.

  3. 03Now

    Look ahead

    Find the past days that looked like today and check in Ask how the day after went.

  4. 04Now

    Try it and check

    After changing a habit, compare before and after to see whether it worked for you.

  5. 05Next

    Judge it before you change it

    Run the options through a model built from your own records and compare the outcomes first. This is what people call a digital twin, and it is the next step Decis is working toward.

Getting started

Gather your scattered records, then read them with the tool you prefer.

Ask inside Decis, or read the standard datasets from the AI you already use or your own scripts.

  1. Connect your sources

    Connect Apple Health, Google Health, WHOOP, and Google Calendar. When the same record arrives by more than one route, Decis resolves the copies before adding anything up.

  2. Connect MCP, CLI, or API

    Paste the MCP address into Claude or ChatGPT, or install the Decis CLI. Explore and query 80+ datasets, read-only.

  3. Review an experiment

    Ask Decis to compare the periods before and after a habit change. It tells you which dates it checked and which records were missing.

Start free

Ask your years of records a first question.

No credit card required. The Free plan includes 25 questions each month.

Connect your records freeView pricing
Answers that cannot reach a conclusion because evidence is insufficient do not use your allowance.