Predictive health platform

A live feed
of your body.

Every meal you eat, every night you sleep, every blood panel you get — flowing into one continuously updating model of you, maintained by a team of AI agents. fiveplate connects what you consume to what your labs say, and flags what's coming before it arrives.

  • Free to start
  • Your data stays yours
  • Works with Apple Health
Live · Body feed
Day 214 of tracking
Resting HR 58 bpmHRV 74 ms ↑

7h 12m

Apple Health

2,340

kcal today

9,412

82 min active

Intake · 3 meals logged
1,780 kcal
Protein
96/140g
Fiber
9/30g
Carbs
182/220g
Fat
54/70g
Lipid panel · Q2 upload

148 mg/dL LDL

above optimal <100 · rising 3 quarters

Forecast · 7-day pattern

Fiber has averaged 9g/day for 7 days — a third of your target. With LDL already rising, this pattern projects further drift by your next panel. Two legume-based meals this week reverses the trend.

LDL cholesterolSleep debtProtein intakeHDLResting heart rateFiberHbA1cCalorie burnTriglyceridesHRVHemoglobinSodiumVitamin DActive minutesFasting glucoseCaffeineIronDeep sleepSugarStepsLDL cholesterolSleep debtProtein intakeHDLResting heart rateFiberHbA1cCalorie burnTriglyceridesHRVHemoglobinSodiumVitamin DActive minutesFasting glucoseCaffeineIronDeep sleepSugarSteps

Every signal your body emits, in one model

The model

Four streams in. Foresight out.

Trackers show you what you did. A model of your body tells you what happens next.

  1. 01

    Log what you eat

    Point your camera at any plate — AI names the dish and fills in calories, protein, carbs, fat and fiber against a real serving. Every meal becomes a data point, not a chore.

  2. 02

    Connect your wearable

    Sleep, calorie burn, steps, heart rate — pulled from Apple Health or whatever device you wear. The model sees your output, not just your input.

  3. 03

    Upload your bloodwork

    Drop in your quarterly panel — lipid profile, HbA1c, hemoglobin, vitamins. fiveplate parses it, tracks every marker against its optimal range, and remembers your history.

  4. 04

    The model watches

    This is the point of all of it. Patterns across meals, sleep and labs become forecasts: what's drifting, why, and what to change before it shows up in your next report.

Agents

Five agents orbit every plate

It's in the name — and on the logo. fiveplate isn't one model answering one prompt; it's a team of specialist AI agents, coordinated by an orchestrator, working your data the moment it lands.

fiveplate

The five nodes on the plate are the five agents: Orchestrator, Logger, Physician, Forecaster and Feed.

Agent trace · one logged meal
12:41:03orchestratormeal.photo received → routing to logger + physician
12:41:04loggerdish: rajma chawal · 520 kcal · 22P/68C/14F · fiber 11g → written
12:41:05physicianfiber +11g vs 7-day deficit · LDL history checked → drift risk easing
12:41:05forecasterweekly fiber projection ↑ 18g → 24g · next-panel outlook updated
12:41:06feedpost composed → published · competition scores recomputed
12:41:06orchestratordone in 2.8s · model of you updated

Orchestrator

Routes every event — a photo, a synced night of sleep, an uploaded panel — to the right specialists, and merges their answers into one picture of you.

Logger

The vision specialist. Names the dish from a photo, sizes the serving, writes calories, protein, carbs, fat and fiber into your record.

Physician

Reads every new data point against your biomarker history and reference ranges — what 148 LDL means, what it risks, what to change.

Forecaster

Projects your trajectories. Recomputes where fiber, lipids and recovery are heading every time new data lands — and raises flags early.

Feed

Runs the social layer — composes your posts, scores competitions, keeps streaks honest — without ever touching your private health data.

New specialists join the orbit as fiveplate grows —
a labs parser, a sleep coach, a training planner.

Your data

The most complete picture of you, anywhere

Your doctor sees you twice a year. Your watch sees your wrist. fiveplate sees the whole system — and what each part does to the others.

The core

From tracking to prediction

Numbers in isolation are trivia. Connected, they become foresight. Seven days without fiber is a note in a diary — seven days without fiber plus a rising LDL history is a forecast, and fiveplate makes it before your next panel does.

Observed

Fiber averaged 9g/day for 7 days. LDL rose 118 → 148 mg/dL across your last three quarterly panels.

Forecast

On this trajectory, LDL stays above the optimal <100 range at your next panel — the zone where cardiovascular risk starts compounding.

Action

Two legume-based meals a week and 25g+ daily fiber reverses the drift. fiveplate tracks the correction meal by meal.

Meals, from a photo

AI reads calories, protein, carbs, fat and fiber off the plate — home food, street food, restaurant plates. No database searching.

Bloodwork, decoded

Upload a panel and every marker lands against its optimal range — what 148 LDL means, where it should be, and what it risks if it stays.

Your wearable's stream

Sleep stages, calorie burn, steps and heart rate flow in from Apple Health — the output side of your body's ledger.

Trends, not snapshots

A single reading is noise. fiveplate keeps your history so every marker becomes a direction — improving, holding, or drifting.

Goals that do the math

Target weight, timeline, and now your labs — daily calories and macros calibrated to where your body actually is.

Yours, and only yours

Health data is the most personal data there is. Everything is private by default — you choose what, if anything, ever reaches the feed.

Foresight

Prevention, almost predicted

When one model holds your meals, sleep, movement and blood history, problems announce themselves early. These are the calls it exists to make.

Lipids + diet

History of high cholesterol, and the last two weeks of meals ran heavy on saturated fat. fiveplate flags the collision course — and the exact meals driving it — months before your next panel would.

Sleep + intake

Five short nights in a row, and your logged intake always spikes the day after. The model surfaces the loop — poor sleep is eating your calorie target — so you can break it at the cause.

Burn + fuel

Training load is up 40% but protein hasn't moved. Before it costs you recovery, fiveplate points at the gap and sizes it: 35g a day short, here's what covers it.

Compounding

Every day you log, the model knows you better

A week in, it knows your macros. A quarter in, it holds your meals against a blood panel. A year in, it has seen how your body responds to food, sleep and training across seasons — a depth of personal data no annual checkup comes close to.

  • Week 1Your intake baseline: calories, macros, fiber, caffeine.
  • Month 1Your rhythms: sleep vs. appetite, training vs. recovery.
  • Quarter 1Your first panel lands — diet and labs finally in one frame.
  • Year 1Forecasts personal to your physiology, not population averages.
Start compounding

Adherence

A model is only as good as the data you feed it

So fiveplate makes feeding it the easiest habit you have — streaks, friends, and a feed that makes logging feel like posting.

Streaks keep the feed alive

One log a day keeps the flame — and an unbroken data stream is exactly what makes predictions sharp.

Compete on real signals

Weekly boards on protein, meals and streaks against friends or your whole city. Healthy competition, feeding a serious dataset.

Social is optional

The feed and boards exist to keep you logging. Every meal can stay private — the model works for you either way.

FAQ

Questions worth asking

What exactly does fiveplate collect?+

What you choose to give it: meals (photo or typed), coffee, sleep and activity from Apple Health or your wearable, and the blood panels you upload — lipid profile, HbA1c, hemoglobin, vitamins and more. Each stream is optional; every stream you add makes the model of your body sharper.

What runs under the hood?+

A team of specialist AI agents — the five nodes on the logo. An orchestrator routes every event to a logger agent (reads dishes from photos), a physician agent (checks new data against your biomarker history and reference ranges), a forecaster agent (recomputes your trajectories), and a feed agent (runs the social layer). One log fans out to the whole team and comes back as an updated model of you.

What does it actually predict?+

Directions, not diagnoses. It connects your consumption patterns to your biomarker history and flags trajectories early — fiber consistently under target while LDL trends up, sleep loss driving intake spikes, protein lagging a training load. Each flag comes with the pattern behind it and a concrete correction.

Is this medical advice?+

No. fiveplate surfaces patterns in your own data and explains standard reference ranges — it doesn't diagnose or treat anything. Think of it as the world's most complete briefing to bring to your doctor, not a replacement for one.

How accurate is the meal AI?+

It's an estimate and says so — every analysis carries a confidence level, and you can correct any number in two taps. Corrections are remembered, so your regular dishes get sharper over time. Indian home cooking, street food and restaurant plates are all first-class.

Who can see my health data?+

You. Meals are public or private per-log, and bloodwork, sleep and biometrics never appear in any feed, leaderboard or competition. The social layer runs on what you choose to share; the model runs on everything, privately.

What does it cost?+

Nothing to start. Sign up, start your feed, connect your data — no card required.

Live · from your first log

Your body is already generating the data. Start listening.

One photo starts the feed. Every log after it compounds into a model that sees around corners.

Start your body's feed

Takes under a minute · no card required