How to Get the Most Out of FuelTron

FuelTron has a lot in it. That's by design. Your health isn't one number, and I didn't want to build an app that pretended it was. But that also means there are features you might not find just by tapping around. This post is a walkthrough of the pieces I think matter most, how they work under the hood, and how to get the best results out of them.

I'm writing this as the developer. If something here is wrong, or you want a feature that isn't covered, email me at contact@decisivelabsllc.com. I read everything.

1. Your FuelTron Score: What It Means

The FuelTron Score is the single number on your Dashboard that tries to answer "how am I doing overall." It's not the only thing to look at, but it's the fastest way to see if you're on or off track.

Under the hood, it's a weighted blend of two sub-scores:

• Nutrition Score (55% weight). How well you're eating today and across the last week.
• Health Score (45% weight). How your body is actually responding, based on bloodwork, Garmin recovery and activity data, and your body composition and strength.

If you haven't connected Garmin or uploaded bloodwork yet and you don't have body comp data in the app, the FuelTron Score falls back to just the Nutrition Score so you still see something meaningful from day one.

How the Nutrition Score is built

The Nutrition Score is made of three parts:

• Calorie adherence (40%). Hitting your calorie target inside a tight band (roughly 95 to 105%) scores highest. The further you drift from target, the lower this component gets.
• Macro balance (35%). How well your protein, carbs, and fat match your goals. Protein gets the heaviest weight inside this component because it's the one most people under-hit.
• Micronutrient profile (25%). Nutrient density, adequacy against RDA for the key nutrients, an anti-inflammatory proxy, and a protein-to-calorie read.

The score uses a 7-day rolling average when you have at least one logged day in that window. This is intentional. One off day shouldn't tank your score, and one perfect day shouldn't hide a bad week.

How the Health Score is built

The Health Score blends up to four pillars, each contributing equally by default at 25%:

• Biometric recovery. Sleep quality, HRV, resting heart rate, and body battery from a connected Garmin device. This tells you how your body is actually recovering.
• Activity. Steps, movement, and exercise intensity measured against WHO weekly targets. This tells you how much you're actually moving.
• Bloodwork. A biomarker assessment across lipids, glucose, thyroid, hormones, liver, kidney, and more. This pillar's weight is automatically adjusted by how fresh your labs are (see below).
• Body Score. Your FuelTron Body Score, which blends body composition and strength (covered in the next section). This pillar's weight is adjusted by the trust of your body comp data source.

If you're missing a pillar, the remaining ones reweight automatically so your score still reflects the data you do have. If all four are missing, the FuelTron Score falls back to just your Nutrition Score.

Bloodwork decays with age. Labs from the last 90 days count at full weight. At 180 days, about 85%. At a year, 65%. Past a year, 40%. Upload fresh labs when you have them.

Data Confidence and what to act on

Next to your score you'll see a Data Confidence badge. High means you have several logged days, Garmin connected, recent labs, and body composition data in the app. Medium means you have partial coverage. Low means you haven't given the app enough to work with yet. A low-confidence 85 is less meaningful than a high-confidence 75.

For action thresholds, I think about it like this: 80+ is strong, 70 to 79 is fine but worth reviewing, and anything under 60 in any component is worth clicking into to see what's dragging it down. The Dashboard breaks every component out so you can see whether it's calories, macros, micros, bloodwork, recovery, activity, or your body score doing the damage.

The Micronutrient Dashboard card

Your Dashboard also shows a dedicated Micronutrient card that visualizes your 7-day averages for the key nutrients against the RDA. Each nutrient gets a horizontal range bar, green when you're in range, amber or orange when you're drifting low, and red when you're meaningfully under target. This is where you look if you want to know whether you're actually getting enough vitamin D, magnesium, B12, fiber, and the other nutrients that are easy to miss. It's more useful than a table of raw numbers because it shows you which nutrients to act on instead of making you interpret them yourself.

The Heart Health Score

There's also a Heart Health Score card on the Dashboard. It's a 0 to 100 number that looks at your dietary fat patterns from a cardiac-risk angle, anchored in modern cardiology consensus from 2015 onward rather than the older dogmas about dietary cholesterol and the omega-6 ratio. Inputs are your saturated fat, trans fat, omega-3, omega-6, fat quality (monounsaturated plus polyunsaturated vs total), and dietary cholesterol.

The biggest drivers of the score are saturated fat, trans fat, and omega-3 adequacy, because those are the strongest cardiac risk signals in the current evidence base. Dietary cholesterol and the omega-6 ratio get small contributions because the evidence no longer supports treating them as primary concerns. The card includes a 7-day trend chart so you can see whether your fat quality is moving in the right direction. If your LDL is elevated and you're working to bring it down with dietary changes, this is the Dashboard card to watch day to day.

2. The FuelTron Body Score (Beyond BMI)

BMI is a bad metric. It can't tell a 190-pound bodybuilder at 10% body fat from a 190-pound office worker at 30% body fat, because it doesn't know about muscle or strength. The FuelTron Body Score is my replacement. It's a 0 to 100 number that shows up on its own Dashboard card and feeds your Health Score as one of the four pillars, weighted by a trust factor so a DEXA reading counts more than a tape measurement.

What goes into it

• Body fat percentile (40%) against age and sex norms.
• Lean Mass Index / FFMI (25%), your lean mass divided by height squared, which rewards muscle.
• Relative strength (25%) based on your lifts vs age- and sex-adjusted percentile tables.
• BMI sanity check (10%) to catch extreme outliers.

If you're missing body comp or strength data, the remaining components reweight. If you have neither, the score falls back to BMI alone and is capped at 70, because BMI on its own can't distinguish a lean athlete from a sedentary person.

How to enter your data

Both body composition and strength go in through Profile, then Targets and Body. For body composition, FuelTron uses a trust hierarchy: lab sources (DEXA, BodPod, hydrostatic) count at full weight, smart scales (Garmin Index, Withings, Renpho, and similar) at 90%, and self-report (Navy tape, calipers, visual estimate) at 70%. A built-in Navy tape calculator is there if you don't have a scale or a recent lab.

For strength, enter a recent working set for the four headline lifts (bench, squat, deadlift, overhead press) in the format of weight times reps at your bodyweight. The app estimates your one-rep max via the Epley formula and compares it to age- and sex-adjusted percentile tables. A hard set of 5 to 10 reps is plenty. No need to actually max out.

Work in progress

The Body Score is newer than most of what's in this guide. The component weights, percentile tables, and trust factors will continue to evolve. If you use it and the number feels off for your situation, tell me. That's how it gets better.

3. AI Photo Logging (And How to Correct It)

The single most-used feature in FuelTron is AI photo logging. Point your camera at a meal, tap, and FuelTron returns an estimated breakdown. That sounds simple. The reality is that getting photo logging right is one of the hardest problems in nutrition apps, and I want to walk through how FuelTron approaches it.

One quick note on the camera itself. FuelTron uses a custom in-app camera, not a handoff to your phone's default camera app. You'll see corner brackets, a live preview, and a Food versus Label toggle on the shutter (Food is the wide bracket layout for meals, Label is the tall bracket layout for supplement bottles and bloodwork pages). The reason for the custom camera is framing: the shape of the brackets nudges you to line the shot up the way the AI wants to see it, which improves accuracy before the image ever leaves your phone.

How the AI typically fails

Most AI food photo tools fail in a few predictable ways:

• They guess a single number (say, "680 calories") with false confidence, giving you no way to know which parts are solid and which are guesses.
• They underestimate portion sizes, especially for medium and large meals. This is well-documented in the research literature.
• They can't reason about 3D volume from a 2D photo, so they rely on averages.
• They confuse similar-looking ingredients (full-fat yogurt vs non-fat, for example) even though those can shift calories by 20% or more.
• They hallucinate vitamin and mineral values in the wrong units, mixing up milligrams, micrograms, and IU.

What FuelTron does differently

The photo logging prompt is structured in stages. First it identifies the visible components (chicken, rice, broccoli) as separate items rather than lumping everything into one blob. Then it reasons about portion size using visual anchors like plate diameter and utensils. Then it estimates the cooking method, because grilled vs fried is a big delta. Only then does it compute nutrition.

On top of that, the AI runs a confidence check on each component. If an assumption would materially change the result (full-fat vs non-fat, grilled vs breaded, 6oz vs 10oz), it flags it and shows you the alternatives. Things that don't matter (organic vs conventional blueberries) it ignores.

Finally, after the AI returns a result, FuelTron runs deterministic sanitization on the back end. If vitamin D comes back over 50 mcg, the system assumes the model confused IU with mcg and converts it. Vitamin A gets the same treatment. Omega-3 and omega-6 values are checked against the total fat ceiling. This catches an entire category of hallucinated numbers before you ever see them.

The angle sensor: aim for 45°

Vision models struggle to reason about 3D volume from a 2D photo, which is a root cause of portion-size underestimation. FuelTron's food camera has a live angle sensor that fixes this at the source. Research on AI food portion estimation consistently shows vision models estimate volume much more accurately from a 45° oblique angle than from a top-down or straight-on shot, because at 45° the model can see height and shape, not just an outline.

Open the food camera and you'll see a live angle readout. It turns green and shows "ANGLE ✓" in the optimal 38 to 55° zone, amber when you're close, and gives contextual guidance ("Raise angle, tilt phone back" or "Lower angle, tilt toward plate") when you're off. Fixing the input before the AI runs is the cheapest accuracy improvement possible.

How to correct the AI

When the AI gets something wrong, you don't retake the photo. You just talk to it. After the initial analysis, your chat bar enters "editing meal" mode and every message you send is treated as a correction. For example:

• "The chicken was closer to 8oz, not 5."
• "That's balsamic glaze, not gravy."
• "I forgot to mention there's about a tablespoon of olive oil on the broccoli."
• "The rice is brown, not white."

Each correction sends the current meal JSON plus your note back to the model, which returns a recalculated meal. You can do this up to five times per photo before the app suggests retaking the shot. Most of the time one or two corrections is enough.

What the AI remembers from your corrections

Corrections aren't just applied to the current meal. When a correction is significant and generalizable ("I always use non-fat Greek yogurt" or "I'm lactose intolerant"), the app extracts it as a memory after the conversation ends and stores it so future meals benefit from the same context. More on the memory system further down.

4. Supplement Label Scanning

Supplements are tedious to log manually because the label has a dozen numbers on it and you have to get the units right. FuelTron handles this with a vision model that reads the Supplement Facts panel.

How it works

Open the supplement logger, point your camera at the label, and the AI extracts the name, brand, serving size, servings per container, each active ingredient with its dosage, and the unit. You review the extracted values, fix anything that looks off, and save it to your personal supplement database. From then on, logging that supplement is one tap.

If you have several bottles lined up and you get them into a single shot, the extractor will pick up each one and return them as separate supplements. This is the fastest way to onboard a multi-supplement stack. Stand the bottles in good light, get the labels facing the camera, and fire.

Always double check the units

This is the most important thing I can tell you about supplement logging. Vision models are excellent at reading text but occasionally confuse mg (milligrams) and mcg (micrograms), which are 1000x apart. For vitamin D in particular, some labels list it in IU and the model has to convert. FuelTron has back-end guards that catch the obvious errors, but you are the last line of defense. When the review screen comes up after a scan, glance at each unit and make sure it matches the label. This takes two seconds and it's worth doing every single time.

What gets tracked

Every food and supplement entry in FuelTron carries 34 mandatory nutrients, including the big macros, eight fat subtypes, cholesterol, eight key minerals, 13 vitamins, plus bioactives like polyphenols, curcumin, EGCG, resveratrol, anthocyanins, beta-carotene, lycopene, and lutein. This is intentionally more than what any standard tracker covers because bioactive compounds from whole foods and specific supplements matter for long-term health even if they don't show up on a traditional nutrition label.

What isn't tracked (yet) and how to request something

A few specialty compounds and newer supplements aren't on the mandatory list yet. If you take something you want tracked, email me at contact@decisivelabsllc.com with the name and any research you want me to look at. I add compounds when I can validate the units, RDA or AI, and the health relevance. The fastest path to adding a nutrient is a direct user request with good source material.

5. My Foods: The Speed Lane

Photo logging is the main input for new foods. My Foods is what you use once you've already logged something and you just want it back tomorrow.

Any food or meal you log can be saved to My Foods with a tap. After that, it shows up in the food log search bar with a green indicator. Tapping a green result adds it instantly, with no AI call, no waiting, and full nutrition preserved. All 34 nutrients carry over. This matters if you eat similar breakfasts most days, rotate through a few lunch options, or have a go-to post-workout meal.

You can also save multi-food meals (your "Sunday Breakfast Bowl" with eggs, oats, berries, and yogurt) as a single entry. Log the whole thing as one tap.

There's a My Foods page in the app where you can see everything you've saved, sorted by how often you've logged it. Your most-used foods float to the top automatically, so the speed lane gets faster the more you use it. If you copy a meal from the social feed, the same save option is available there, with full nutrition preserved.

6. Calories and Macro Balance

Your basal metabolic rate

When you onboard, FuelTron calculates your initial calorie target using the Mifflin-St Jeor formula, which is the current standard equation for basal metabolic rate. Your BMR is the number of calories your body burns at complete rest, just to keep you alive. It's what it costs to run your heart, your lungs, your brain, and every other system when you're doing nothing at all. Think of it as the floor of your daily calorie need before any movement gets added.

The formula takes your height, weight, age, and biological sex and returns a BMR value. From there, FuelTron multiplies that BMR by an activity multiplier that accounts for everything you do beyond lying on the couch. That final number is your estimated Total Daily Energy Expenditure, or TDEE, and it's what your daily calorie target is built from.

Activity level definitions

This is the setting in your profile where the most common user mistake happens, and it's worth slowing down on. FuelTron uses five activity levels, each tied to a specific multiplier:

• Sedentary (×1.2). Little or no exercise. This is most office workers, remote workers, students, and anyone whose day is mostly sitting. If you walk the dog once a day and do some chores but you don't do structured exercise, you're sedentary.
• Lightly Active (×1.375). Light exercise 1 to 3 days per week. A couple of gym sessions, a few walks, an occasional bike ride.
• Moderately Active (×1.55). Moderate exercise 3 to 5 days per week. Consistent structured training most days. Lifting, running, cycling, classes.
• Very Active (×1.725). Hard exercise 6 to 7 days per week. You train almost every day and you train hard.
• Extremely Active (×1.9). Very hard daily exercise plus a physically demanding job. Construction, warehouse work, farming, or two-a-day athletic training.

Most people overestimate their activity level

This is the single biggest reason I see people track their food honestly and still fail to hit their weight goals. They pick Moderately Active or Very Active because they feel active. They're on their feet some of the day, they hit the gym a few times a week, they're "not lazy." So they set the multiplier at 1.55 or higher, the app calculates their calorie target off of that, and then they eat to that target and wonder why the scale isn't moving.

The honest answer is that most people reading this are probably Sedentary or Lightly Active, and the gap between Lightly Active (1.375) and Very Active (1.725) on a 1700-calorie BMR is over 600 calories a day. That's enough to erase a calorie deficit entirely, or worse, put you in a surplus while you think you're cutting. If your weight isn't going where you want it to, the first thing I'd question is your activity level setting, not your food logging.

A useful rule of thumb: if your exercise isn't structured, isn't most days of the week, and doesn't make you sweat or breathe hard, assume you're one notch lower than you think. You can always move up if the scale tells you to.

Macro distribution

Once FuelTron has your calorie target, macros are distributed based on your goal:

• Protein defaults to roughly 1.6 to 2.0 grams per kilogram of body weight, which is the evidence-backed range for people who are active or trying to preserve muscle while losing fat.
• Fat defaults to about 25% of calories (higher for low-carb profiles).
• Carbs fill the remainder.

The sliders

Everyone's body is different, and the defaults are a starting point, not a prescription. In Profile, Settings and Help, you can enable Custom Macro Preferences and use sliders to adjust protein, carbs, and fat by up to 50% in either direction. The app recalculates your daily calorie target from the new macro split and keeps it consistent everywhere (Dashboard, Food Log, Profile).

If you're running a high-protein cut, nudge protein up. If you train fasted and need lower carbs in the morning, skew carbs down. The sliders are there because I don't know your body better than you do.

What's coming

A few related features are on the roadmap but not yet shipped:

• Direct micronutrient goal adjustment (setting your own targets for specific vitamins or minerals).
• Bloodwork-informed suggestions, so when you upload labs the app can recommend specific macro and micro adjustments based on what your markers say.
• Profile factors like blood type as optional inputs.

These are under active consideration. If one of them is important to you, tell me which one and why.

7. Exercise, Routines, and Calorie Burn

Exercise logging in FuelTron has three flavors, and knowing the difference will save you time.

• Sessions. One-off entries. Good for a spontaneous hike or a pickup basketball game.
• Templates. Saved workout structures you want to reuse. Your "Push Day" template with the usual sets and reps. Load it anytime, adjust for the current session, save.
• Routines. Templates with a schedule attached. Your "Monday/Wednesday/Friday Legs" routine shows up on the right days automatically.

There are also pre-built quick-start templates (Push Day, Pull Day, Leg Day, Run, Swim, HIIT) if you want a sensible default to start from.

How calorie burn is calculated

This is the part that most apps get wrong and I care a lot about. FuelTron calculates exercise calories using the 2024 Ainsworth Compendium of Physical Activities, which is the current gold standard for MET values across activities. The formula is:

calories = MET × kilograms × hours × demographic factor + elevation bonus

MET values come from a lookup table of 23 activities at four intensity levels each (low, moderate, high, very high), so "running at 6 mph" and "running intervals" score differently. The demographic factor adjusts for known physiology differences (female users get a small downward adjustment, and users 40 and up get a small downward adjustment reflecting average metabolic differences). For hiking, running, and walking with elevation gain, an elevation bonus is added based on the actual meters climbed.

Importantly, this is completely deterministic. No AI call, no variability, no surprise numbers. Two identical workouts produce identical calorie burns.

How to adjust it

If you know your watch or a lab test shows you burn more or less than the formula estimates, you can adjust. Every exercise card has an inline edit, and when you save, you get two options: Save Update (applies your number to this entry) or Save as New (creates a new version without overwriting the original). Next to the calorie number there's a small info icon that opens a breakdown popover showing exactly which MET value was used, the weight factor, the demographic factor, and the elevation bonus. This is so you can see why the app gave you that number and decide if you want to override it.

The AI Coach Review

After every new or significantly-updated exercise entry, FuelTron runs an AI Coach Review in the background. This looks at your exercise in the context of today's nutrition, your other activity, last night's sleep, your weekly targets, and your goal direction. It flags entries that look over- or under-estimated and notes the impact on your calorie balance, protein needs, and recovery. You'll see the result on the exercise card when it's done. It doesn't block anything. It's there to catch obvious mistakes and give you a second set of eyes.

8. Garmin Connect and Sync

If you have a Garmin watch, connecting it is the single biggest upgrade you can give FuelTron. Go to Profile, Connected Devices, tap Connect Garmin, and authorize the app on Garmin's page. FuelTron uses OAuth 2.0 with PKCE, and your tokens are encrypted at rest. When the connection finishes, the app backfills 30 days of history so you don't start from empty.

After that, FuelTron auto-syncs every time you open the app, so you never have to think about it.

What data flows in

• Daily summaries: steps, distance, calories burned, resting heart rate, stress, body battery.
• Sleep: duration, stages (deep, light, REM, awake), sleep score, SpO2.
• Activities: type, duration, distance, calories, heart rate zones.
• Epoch data: 15-minute slices of steps, heart rate, MET values, and intensity.
• HRV: weekly average, last night, HRV status vs your baseline.
• Body composition: weight, BMI, body fat, muscle mass (from connected scales).

The Garmin and Fitness pages

The Fitness tab surfaces your daily summary, a timeline of your activity through the day, 7-day trends, and weekly activity against the WHO targets (150 minutes moderate or 75 vigorous, plus two strength sessions).

The Garmin page is where the correlations live. FuelTron looks across your nutrition, sleep, stress, and activity data and surfaces seven correlation types:

1. Calorie Balance. Garmin-measured burn vs FuelTron-logged intake, day by day.
2. Post-Meal Response. How your heart rate responds in the 15 minutes after a meal, correlated with carbs and total calories.
3. Sleep-Nutrition. How your sleep score relates to what you ate in the evening.
4. Stress-Eating. How stress level and body battery track with meal timing and content.
5. Exercise Recovery. How your post-workout meals line up with what you just did.
6. Resting HR vs Diet. How your resting heart rate trends with dietary changes.
7. HRV vs Diet. How overnight HRV relates to alcohol and sugar intake the day before.

These aren't generic charts. They're computed from your data, so the patterns you see are yours. This is where the "whole system" view of health starts to feel real.

9. Bloodwork Upload and Trends

Bloodwork is one of the most valuable inputs you can give FuelTron. It's what turns a nutrition tracker into a health tracker. There are three ways to get labs into the app, and they're not all equally reliable.

The best way: PDF directly from your lab's portal

If you get your labs from Quest, LabCorp, or any major provider, they almost always make a PDF available in your patient portal. Download that PDF and upload it directly to FuelTron. This is the method I've put the most work into making reliable.

The extractor is a three-stage pipeline. It first reads the PDF text with a medical lab parsing prompt tuned for standard lab formats. If the AI pass comes back weak, it falls back to regex pattern matching. And for scanned or image-based PDFs where there's no extractable text, it automatically switches to vision-based extraction. The review screen shows you a confidence score on the extraction so you can see at a glance whether the result is solid before you save it.

I test this flow regularly with real Quest and LabCorp PDFs, which is why I recommend this path first. The more standard your lab's report layout, the cleaner the extraction.

Photos and images also work (but with more errors)

If you only have a printed paper report, you can take a photo with your camera and upload it. Same goes for a screenshot or image file from your photo library. These flow through the vision model, and they do work, but the error rate is higher than a direct PDF upload. Lighting, glare, cropping, and print quality all add noise the model has to work through. Use photos when you don't have a PDF option, not as the default.

Manual entry

You can also type your values in by hand. Tap through to manual entry, set the test date, optionally add the lab name and ordering doctor, and then add each marker with its name, value, unit, and reference range. Manual entry is a good option if you only want to track a handful of markers, or if you're working from a report that the extractor struggles with.

Review before you save

Every upload method, whether PDF, photo, or manual, goes through a review step before anything lands in your record. You'll see each extracted marker with its value, unit, reference range, and in-range status. Anything that looks wrong can be edited directly in the review screen. Double check any value where the unit matters (vitamin D, B12, and thyroid markers in particular). You're always the last set of eyes before the data is saved.

Privacy

PDFs and photos are processed in memory only. The original file is never written to disk or stored on a server. Only the extracted values and the test metadata (test date, lab name, doctor name if you entered one) are saved to your account. If the thought of a bloodwork app holding onto your raw PDF made you hesitate, it shouldn't here.

What you see after upload

Your bloodwork shows up on the Bloodwork page, organized into nine categories: lipids, glucose and metabolic, liver, kidney, thyroid, hormones, vitamins, minerals, and other. Each marker shows the latest value with a color-coded status dot (green for in-range, amber or orange for drifting, red for critical or far out of range) and a trend arrow when the value has changed meaningfully since your last test.

FuelTron supports more than 80 biomarkers across the standard panels, plus a lot of specialty markers like ApoB, Lp(a), homocysteine, hs-CRP, the full thyroid panel, the full iron panel, and more. If your lab ran it, there's a good chance FuelTron knows about it.

Tracking trends over time

This is the payoff for logging bloodwork consistently. Tap any marker and FuelTron opens a detail view with a large current value, a horizontal range bar showing where you sit relative to the reference range, and a trend line chart built from every historical upload of that marker. Two data points is enough to see a direction. Five or six and you can tell whether your dietary changes are actually moving the numbers.

As a concrete example: if you're working to lower LDL, upload each set of labs as they come back. The trend line for LDL will show you the slope. If it's coming down, you know what you're doing is working. If it's flat or climbing despite your changes, that's useful information too. The whole point of tracking is to close that feedback loop.

Edit and delete anytime

You can edit an existing bloodwork entry or delete it entirely from the Bloodwork page. If the extractor got something wrong and you didn't catch it in review, or if you entered a test twice by accident, you have full control. There's no penalty for cleaning things up.

How bloodwork affects your FuelTron Score

I covered this in the scoring section, but it's worth repeating here because it changes how often you should upload. Bloodwork is one of the four pillars of your Health Score, and its weight is adjusted by how fresh your labs are. Labs within 90 days count at full weight, about 85% from 90 to 180 days, 65% through a year, then 40% after that. Uploading new labs when you get them is the fastest way to keep that pillar contributing accurately to your score.

10. The AI Assistant (What It Can and Can't Do)

The AI Assistant is the chat interface at the bottom of the app. It's available from nearly every screen, and it's designed to be the fastest way to get something done.

What it can help with

• Logging food by description ("I had two eggs and a slice of whole wheat toast with butter").
• Logging exercise by description ("I ran 3 miles at an easy pace").
• Explaining what a bloodwork result means in context of your nutrition.
• Looking for correlations across your logs, bloodwork, and activity.
• Telling you which nutrients you're low in this week or might be overdoing.
• Suggesting foods that would help with a specific goal.
• Recommending exercises based on what you've done recently and how you're recovering.
• Updating your profile and goals when you ask it to.

What it's not for

The AI Assistant is not a general-purpose chatbot. It won't help with your mental health, relationship advice, coding problems, or anything else outside of nutrition, fitness, health, and activity. This is a deliberate design choice, enforced in the code with a semantic firewall. FuelTron is also not a substitute for medical advice. When bloodwork or symptoms suggest a real medical issue, the assistant will tell you to consult a professional, and that's the right answer.

The hidden recipe feature

This is one of my favorite things in the app and almost nobody finds it on their own. Open the assistant and type something like "What should I eat for lunch?" or "Suggest a high-protein dinner." FuelTron will detect the intent and return a set of meal suggestions tailored to your remaining macros, goals, and preferences.

Pick one and ask for the detailed recipe. The assistant will return a full recipe with ingredients, instructions, prep time, and nutrition. You can modify it on the fly ("make it lower carb" or "swap the chicken for tofu") and the recipe will be updated while keeping everything else intact. When you're happy, log it directly from the recipe view and the full meal lands in your food log with all 34 nutrients populated.

This is the closest thing FuelTron has to a personal chef. Use it when you're stuck.

It can log things for you, too

When you ask the assistant to log a food or an exercise, it doesn't just describe what you should do. It actually creates a pending entry with the nutrition or calorie calculation filled in, and shows it to you for review. You approve it, adjust it, or send another correction in natural language. This is much faster than walking through a form, especially for meals you can describe in one sentence.

The same applies to updating your profile. Tell the assistant "I'm going to train twice a day this month, bump my activity level," and it'll make the change.

11. AI Memory

The assistant remembers what matters. After every conversation, FuelTron runs a background extraction that pulls out up to five durable facts and saves them to your personal memory. These are shared between the Assistant and the Coach, so once either one learns something about you, both act on it.

Memories fall into a few categories:

• Preferences. "Prefers non-fat Greek yogurt", "doesn't like seafood".
• Health facts. "Has Type 2 diabetes", "is lactose intolerant".
• Behavior patterns. "Eats most calories in the evening", "skips breakfast on workout days".
• Dietary. "Following a Mediterranean-style diet."
• Lifestyle. "Runs three times per week", "travels for work monthly".
• Communication style. "Prefers concise answers", "likes data over encouragement".

You have full control. Go to Profile, Settings and Help, and open the AI Memory section. You'll see every memory the app has stored about you, grouped by category, each with a delete button. You can remove any memory you don't want, or clear all of them at once. Nothing is hidden. If you want to see exactly what the AI thinks it knows about you, this is where you look.

Most chat AIs forget who you are the moment the session ends. FuelTron carries the context forward so that over time the app becomes more tailored to you, not less.

12. The Social Layer

FuelTron has a built-in social layer if you want one. Connect with friends, share meals and workouts as posts, comment on each other's posts with reactions and inline media, and copy a friend's meal directly into your own food log with all 34 nutrients preserved. It's opt-in and quiet. No algorithmic feed pushing content at you, no strangers in your face. The original idea was for my family to keep each other honest about food and exercise, and the feature grew from there. If that sounds useful, head to the Social tab. If not, ignore it. Friends search is by name only (no email exposure, no "people you may know" suggestions), and you can mute or remove friends any time.

13. Free vs Pro

FuelTron has a free tier and a Pro tier. Pro is $7.99 a month or $69.99 a year, which works out to about $5.83 a month if you go annual. Less than a coffee.

What you get

For that price, you get AI food photo logging, 34-nutrient tracking on every food and supplement, bloodwork PDF upload and trend analysis across 80+ biomarkers, Garmin integration with seven correlation types, MET-based exercise calorie math, an AI Assistant that can log food and generate recipes, a Coach with three personality settings, and a memory system that actually remembers who you are.

MyFitnessPal Premium is $19.99 a month. Noom is closer to $70. Cal AI is $9.99 and only does photo food logging. You'd pay more than FuelTron for a single feature of FuelTron.

The free tier is a real product

The free tier isn't a nagware trial. You can log food, track calories and macros, enter bloodwork manually, log exercise, see your dashboard, use My Foods, connect Garmin, and get your FuelTron Score without ever paying a dollar. Plenty of people can get real value from the free tier alone.

Pro unlocks the features that run on AI every time you use them: the higher-volume photo food logging, bloodwork PDF and photo extraction, unrestricted AI Assistant usage, and the Coach. Frontier vision and language models cost real money per call, and those are the calls that add up. If you're going to use the AI every day, Pro is the tier that makes sense. If you're not, the free tier is ready.

A few closing thoughts

The features in this guide work best when you use them together. The AI photo logger gets smarter when memory knows your preferences. Your macro sliders inform what the meal suggestions pick. Your Garmin data shapes what the Coach says and feeds directly into the Health Score. The FuelTron Score tells you if all of this is actually moving you in the right direction.

If you're new, my suggestion is this: spend the first week logging food with the photo tool and correcting it when it's wrong. Connect Garmin if you have it. Upload bloodwork if you have recent labs. Let the app learn you. By the end of the first week your Data Confidence will be at Medium or High, and the numbers you see on the Dashboard will actually mean something.

And if something is missing, or broken, or confusing, email me. FuelTron gets better because users tell me what they want. That's the whole reason I built it.