Concept site · Phase 0 prototype · nothing here is a live product yet
GlycoTwinPhase 0 Let's get started

Singapore · Founding cohort · 300 places

You already wear the sensor. Now run the experiment.

Pick one habit. Test it fairly for two weeks against your own baseline. Find out if it actually changed your numbers.

Let's get started See how it works

General wellness only. Not a medical device.

A woman scans the glucose sensor on her upper arm with her phone Sweet Life / Unsplash
Walk after lunch · exampleRecovery Improving−55 min

Why it's different

Your baseline, not an average

A week of your normal life sets the bar. Every result is measured against you.

A fair test

On and off days are shuffled, so a good week of sleep can't pose as a good habit.

A straight answer

Improving, no clear effect, or worsening, with how consistent it was. Keep it or drop it.

People walking along a promenade in warm evening light Jonas Weckschmied / Unsplash

Ten minutes of walking after dinner. Does it help you? Two weeks will tell.

Let's get started

Go deeper

How it works

Three weeks. One habit. A fair test.

The method is borrowed from n-of-1 trials, which clinical researchers use to test a treatment on a single person by alternating it on and off. GlycoTwin applies the same idea to everyday habits, and keeps it simple enough to do on a phone.

Experiment · 10-min walk after lunchExample data
5678 060120180 min mmol/L ← 55 min sooner →
Walk days (n=7)No-walk days (n=7)Bands = day-to-day spread
W––WW–W–W––WW–
VerdictImproving
Recovery−55 min
Consistency6 of 7 pairs
A woman at a kitchen counter checks her glucose sensor with her phoneSweet Life / Unsplash

The protocol

Four steps, about three weeks

  1. 1 · 7 days

    Learn your baseline

    Wear your sensor and live normally. We learn your typical meal responses, overnight pattern and day-to-day spread.

  2. 2

    Pick one habit

    Choose from the library. Each experiment names the meal or time window it measures, so the test is fair.

  3. 3 · 14 days

    Follow the shuffled schedule

    The app assigns on and off days in random order. Tap once to log that you did it.

  4. 4

    Get the verdict

    Improving, no clear effect, or worsening, with a consistency score. Keep it, drop it, or run it again.

Experiment library · launch set

Eight habits people already argue about. Find out which ones are true for you.

A customer orders at a brightly lit hawker stall at nightJoylynn Goh / Unsplash
Movement

Walk after meals

10 minutes within 30 minutes of eating. Measures the next three hours.

Order

Vegetables first

Fibre and protein before starch at the same meal. Measures that meal.

Timing

Earlier dinner

Finish eating three hours before bed. Measures overnight stability.

Sleep

Extra 45 minutes

Earlier lights-out, confirmed by your wearable. Measures next-day meals.

Breakfast

Protein-forward start

Swap a carb-led breakfast for a protein-led one. Measures the morning.

Training

Resistance session

A short strength session on on-days. Measures the following 24 hours.

Caffeine

Coffee after food

Move the first coffee after breakfast. Measures the breakfast response.

Local plate

Rice swap

Half the rice, more vegetables at hawker meals. Measures that meal.

A runner checks workout data on a sports watchstreetsh / Unsplash

What we measure

Plain measures, named for what they describe

Glucose Stability Window

How much of your day readings stayed within your own steady band, learned from your baseline rather than a clinical threshold.

Recovery time

Minutes from a meal's peak back to your pre-meal level. The measure most habits move first.

Spike frequency & size

How often readings rose sharply above your baseline, and how large each rise was.

Metabolic Performance Score

A daily wellness estimate that blends glucose stability with sleep, HRV and activity. When your sensor is off, it runs on wearable data alone.

Planned data sources

Apple HealthHealth ConnectDexcom SteloAbbott Libre & LingoOuraWHOOPGarminApple WatchCSV import

Glucose arrives through Apple Health and Health Connect at launch. Direct manufacturer connections depend on partner approval. Brand names belong to their owners; no partnership is implied.

Evidence & limits

What GlycoTwin will not tell you

Two weeks of one person's data can show a pattern. It cannot prove a cause the way a large trial does, and it says nothing about disease.

  • We do not diagnose, treat, cure or prevent any disease.
  • We do not replace a blood test or a doctor's advice.
  • We do not predict diabetes, cognitive decline or lifespan.
  • We do not simulate organs or "biological age".
  • We do not sell your data. Pooled findings are opt-in and anonymised.
  • If you have diabetes or take glucose-lowering medication, follow your care team instead.

Ready to test your first habit?

Let's get started

Pricing · Singapore dollars

Bring your own sensor. Pay only for the lab.

No hardware bundles. Use the glucose sensor and wearable you already have, or buy a sensor wherever you like.

Free
S$0
  • Sync from Apple Health or Health Connect
  • Daily performance score
  • One experiment at a time
Pro · most popular
S$149 / year
  • Or S$19.99 monthly
  • Unlimited experiments and re-runs
  • 30 and 90-day trends
  • Sensor-off mode on wearable data
Performance+
S$249 / year
  • Or S$29.99 monthly
  • Design your own experiments
  • Share results with your coach or dietitian
  • Early features

Founding cohort: the first 300 members get Pro at S$99 for the first year, and help decide which experiments we build next. Claim a place →

Common questions

Do I need a glucose sensor?

For glucose experiments, yes. Sleep and activity experiments run on a ring or watch alone.

Which sensors work?

Any sensor that writes to Apple Health or Health Connect, including Stelo and FreeStyle Libre apps where available.

Can I cancel?

Yes, any time. Annual plans run to the end of the paid year.

Who is it not for?

People with diabetes or on glucose-lowering medication. Follow your care team.

Singapore's skyline and Gardens by the Bay seen from above on a clear dayMark Stoop / Unsplash

About · A BrainBank pilot

Advice is everywhere. Proof that it works for you is rare.

GlycoTwin started from a simple idea: people already wear sensors and already experiment with their habits, but nobody gives them a fair test. We are building that test, with the honesty about its limits that health products usually skip.

GlycoTwin is a pilot venture of BrainBank, a Singapore AI incubator that builds and backs early-stage AI products with local teams.

Founders

Co-founder · Product & AI

Kush Amerasinghe

25 years shipping first versions of AI products. Co-founded Adobe's creative ML team and led Photoshop Neural Filters; built Amazon's first visual agentic AI platform. Nine US patents. Founder of BrainBank.

Co-founder · Partnerships & capital

Larry Yee

Impact investor and the originator of the n-of-1 idea behind GlycoTwin: let people test lifestyle changes on their own wearable data instead of taking advice on trust.

Now hiring · Singapore

Help us build it

Apply via the waitlist

Founding mobile engineers

iOS and Android, HealthKit and Health Connect pipelines.

Data scientists, n-of-1 methods

Within-person statistics, randomised schedules, honest uncertainty.

Growth & content lead

Turn pooled, opt-in findings into stories people share.

Regulatory counsel

Fractional. Wellness-claims review for Singapore, the US and the UK.

Photography from Unsplash by Sweet Life, Jonas Weckschmied, Joylynn Goh, streetsh and Mark Stoop, used under the Unsplash License.

Founding cohort · 300 places

Let's get started.

Leave your details and we'll write when the alpha opens. Founding members get Pro at S$99 for the first year.

Coaches, dietitians, investors, partners and job applicants: use the same form and pick your role.

We use your email only to contact you about GlycoTwin.