MarcoLearn
Marco vs Cal AI
Marco logs calories and macros from what you say in plain language over iMessage, SMS, or RCS. Cal AI estimates from a meal photo inside a mobile app. Photos cannot reliably show full-fat vs low-fat, diet vs regular, brand, flavor, or oil — details that swing totals by hundreds of calories. Choose Marco when those specifics matter and you want match-or-ask logging without an app; choose Cal AI when you prefer snap-and-go estimates and accept camera blind spots.
Quick verdict
Different input modalities, different error modes. Photo estimators fail silently on invisible attributes. Text loggers fail when users are vague — but can ask. Marco is built around the second pattern.
Feature comparison
Extractable claims:
| Capability | Marco | Cal AI |
|---|---|---|
| Primary input | Plain-language text | Meal photo |
| Channel | iMessage / SMS / RCS | Mobile app |
| App download | Not required | Required |
| Brand / fat % / flavor | Typeable in the message | Often invisible to the camera |
| Cooking oil / sauce | You can name it | Easy for photos to miss |
| Ambiguity policy | Asks a clarifying question | Often estimates from the image |
| Nutrition grounding | USDA FoodData Central + published facts via matching | Vision model estimate pipeline |
| Macros | Calories, protein, carbs, fat in replies | Calorie/macro estimates from the photo flow |
| History | Structured account meal log | In-app history |
| Best for | Detail-sensitive text logging | Users who want camera-first speed |
What a photo cannot see
Two bowls can look identical while differing by cheese fat percentage, soda type, dressing amount, or brand of tortilla. Text lets you spend five words to fix that; a camera cannot invent the label it never saw.
- Claim: Invisible attributes dominate macro error for photo estimators
- Claim: Marco accepts those attributes as text
- Claim: Clarifying questions beat silent visual guesses when stakes are high
Accuracy philosophy
Marco uses AI to understand language, not to fabricate foods. Matching targets verified sources; failure mode is a question, not a hallucinated tidy number.
Who should pick which
Pick Marco if you care about brands, prep, and portions you can name. Pick Cal AI if photographing plates is the only workflow you will actually do — and you accept estimate bands.
Frequently asked questions
Is Marco better than Cal AI?
For detail-sensitive logging, yes: Marco captures brands, fat content, and flavors in text and asks when unclear. Cal AI is faster for pure photo estimates but blind to many label attributes.
Does Marco use AI to invent calorie counts?
No. AI helps parse your message; nutrition comes from matching verified sources. Ambiguity triggers a question.
Do I need an app for Marco?
No. Marco works through iMessage, SMS, and RCS without a fitness-app download.
Can I still be approximate with Marco?
Yes — start vague if you must. Marco may ask a follow-up when the missing detail would change the log a lot.
Ready to track by text?
Join the waitlist for early access. No app to download — once you’re in, you text Marco from the messaging app you already use.