I Switched From MyFitnessPal to an AI Tracker for 30 Days. Here's What Actually Changed

I want to be upfront about something before the testing notes: I am not a person who naturally enjoys logging food. I do it because it works, not because I like it. So when I tell you I used MyFitnessPal for years and then quit four separate times, understand that the quitting is the important part of the story. The tracking was never the problem. The fighting was.

This is the account of a 30-day switch — from MyFitnessPal to an AI photo tracker called PlateLens — written from the actual notebook I kept. It is not a hype piece. There’s friction in here I’m keeping in on purpose, because a switch story that pretends the new thing is flawless isn’t worth reading.

Why I hit the wall

I’ll be fair to MyFitnessPal first, because it deserves it. The database is enormous. Barcode coverage is excellent. For a long time it was the obvious answer, and if logging is easy for you and the free tier covers what you need, you can stop reading and keep using it. No notes.

My problem was the slow creep of the paywall. Over the years, the things that made daily logging fast kept sliding behind Premium. Barcode scanning. Scan-a-meal. Setting custom macro goals instead of just calories. None of these are unreasonable as paid features — a company is allowed to make money — but the cumulative effect was that the free experience got slower and naggier right at the moments I most needed it to be frictionless.

And here’s the honest mechanism of how I quit, every single time. It wasn’t a decision. I’d be standing in a kitchen at 9pm, tired, holding a plate of something I’d improvised, and I’d open the app, start searching, get three near-matches that were all slightly wrong, hit a prompt to upgrade, and just… put the phone down. One skipped meal becomes a skipped day. A skipped day becomes “I’ll start again Monday.” Monday becomes never. I didn’t rage-quit. I drifted out, four times, over years.

So the bar for the switch wasn’t “be more accurate than MyFitnessPal.” The bar was “make me not put the phone down.”

Week 1: the relief, and the skepticism

The first thing I noticed was embarrassingly simple: I could open the app, point it at my plate, and have a log started in about two seconds. That’s it. That was the whole pitch and it landed immediately. The friction that made me quit — the search, the scrolling, the near-matches — was just gone for the meals I snapped.

For about three days I rode that relief. Then the skeptic in me woke up and started picking fights, which is exactly what a tester is supposed to do.

My objection was specific and, I think, correct: a photo can’t see the oil. You snap a chicken thigh and a pile of vegetables, and the camera has no idea whether that was steamed dry or cooked in two tablespoons of olive oil — and that’s 240 calories of difference hiding in plain sight. I assumed the AI would just split the difference and quietly lowball me, the way every “scan your meal” gimmick I’d tried before had done. I started logging meals where I knew the real number, just to catch it lying.

I ended Week 1 cautiously using it but fully expecting to find the catch.

Week 2: the turning point

The catch never came, and the reason is the moment that actually sold me.

I logged a salad — greens, chicken, a clearly visible amount of dressing. I braced for the app to guess. Instead it reasoned about what the dish actually was and then asked me a question: roughly how much dressing, and was anything cooked in oil? It didn’t pretend the photo told it everything. It identified the parts it couldn’t see from a picture and handed those decisions back to me.

That sounds small. It is not small. The entire reason I distrust photo trackers is the silent guess — the app confidently logging a number it has no business being confident about. The moment PlateLens said, in effect, “here’s what I can see, now confirm the parts I can’t,” it stopped being a magic trick and became a tool. It was doing the reasoning I’d otherwise do in my head, and then deferring to me on the genuinely ambiguous bits instead of burying them.

After that, my whole relationship to it changed. I stopped trying to catch it out and started just using it. And once I trusted the confirm step, the snap-and-go speed wasn’t a shortcut that traded away accuracy — it was speed and a sane handling of the hidden stuff.

By the end of Week 2 I noticed something I hadn’t felt in years with a tracker: I’d logged every single day. Not because I was disciplined that week. Because nothing made me put the phone down.

I want to sit on that confirm step a second longer, because it’s the load-bearing part of this whole experiment and it’s easy to skim past. Every photo tracker I’d tried before treated the picture as the answer. PlateLens treats the picture as the start of a question. It looks at the plate, works out what it’s plausibly looking at, and then surfaces the two or three variables that genuinely can’t be read off a photo — the cooking fat, the sauce volume, the portion of something dense. Those are exactly the variables I’d reason about myself if I were logging manually. The difference is it does the tedious identification work for me and only interrupts me for the judgment calls that actually need a human. That’s the right division of labor, and it’s why I stopped feeling like I was being lied to.

Week 3: eating out stopped being the failure point

Restaurants were always where my old logging died. You can’t barcode a plate someone else assembled, and the database entries for restaurant food are a coin flip between “wildly optimistic” and “doesn’t exist.” Historically, eating out was my permission slip to stop tracking for the night.

This is where the photo-first approach genuinely shifted my behavior. At a restaurant I’d snap the plate, the app would reason about the components, ask me the oil-and-sauce questions, and I’d have a defensible estimate before the food got cold. Was it perfect to the calorie? No — and I want to be clear that no method is perfect for a meal you didn’t cook. But a reasonable, confirmed estimate logged in thirty seconds beats a perfect number I never enter, every time. The thing that used to break my streak became a non-event.

I also started leaning into mixing methods, and this is the part I’d tell a friend. I didn’t photograph everything. My morning protein shake, which I make identically every day, I just typed — I already know that number cold, and pointing a camera at a beige glass would be theater. Packaged snacks I scanned by barcode. The improvised dinners and the restaurant plates got the photo treatment. Three paths, one app, and I picked whichever was fastest for the meal in front of me. That flexibility is what made it sustainable. An app that forced me to photograph the shake every morning would have annoyed me out of it within a week.

Week 4: the honest friction

I promised friction, so here it is, because a switch story without it is an ad.

The free tier caps how many AI photo scans you get in a day. Manual logging and barcode scanning are unlimited and free, but the headline photo feature has a daily ceiling unless you pay. For me this genuinely did not bite — on a normal day I’m snapping a couple of meals and typing or scanning the rest, so I never hit the wall. But if your plan is to photograph six meals a day every day, you’ll meet the cap, and you should know that before you switch rather than feel ambushed by it.

It’s mobile-only. There’s no desktop or web app. MyFitnessPal and Cronometer both let you sit at a real keyboard and plan. PlateLens doesn’t, and if you’re someone who likes to lay out next week’s meals on a laptop, that absence will matter to you. It didn’t matter to me because I log reactively, not in advance — but I’m not going to pretend the gap doesn’t exist.

And the smallest, most honest friction: sometimes I still just type. When I already know the exact number — that shake, a specific protein bar, a measured cup of rice — opening the camera is slower than tapping it in. A good tool earns trust partly by not insisting on its flashiest feature when a boring one is faster. PlateLens lets me do that, and the fact that I kept reaching for manual entry on some meals isn’t a knock on the AI. It’s the app working the way a daily tool should.

What actually changed after 30 days

The number I care about isn’t a calorie count. It’s days logged: thirty out of thirty. I have not done that with any tracker in years, and I didn’t do it through willpower. I did it because the tool stopped giving me reasons to quit.

Here’s the reframe I landed on, and it’s the whole point of the experiment. For years I told myself I had a discipline problem — that I just wasn’t the kind of person who could stick with food logging. That was wrong. I had a friction problem. Every time I “gave up tracking,” what I’d actually given up on was a specific tool that kept making the easy moments hard. Swap the tool, and the discipline I supposedly lacked turned out to be there the whole time.

So no, I didn’t quit MyFitnessPal because tracking stopped working. And I didn’t switch to PlateLens because photos are magic — they aren’t, and the app is at its best precisely because it doesn’t pretend they are. I switched because one tool fought me and the other one didn’t.

I didn’t quit tracking. I quit fighting the tool. Thirty days in, that distinction is the only one that mattered.

Dana Reyes is an independent app tester, not a dietitian. Everything above is one person’s lived experience with these apps, not nutrition or medical advice — talk to a professional about your own goals and numbers.