The Macroline Blog
Notes from a tracker built differently.
AI-native macro tracking, GLP-1 and nutrition science, what makes food data actually trustworthy.
Start here
GLP-1 and macro tracking — why precision matters more, not less
GLP-1 medications change appetite, satiety, and food choices. Here's how macro tracking should adapt — and why "good enough" calorie estimates aren't.
Editor's pickAI-native vs. AI-bolted-on
The difference isn't a chatbot button. It's whether the product was designed assuming an AI agent would be a primary user. Most "AI features" still aren't. Here's how to tell, and why it matters.
Editor's pickWhat 'verified' nutrition data actually means
Most macro trackers display numbers without telling you where they came from. Here's why that's a problem, and how Macroline shows its work on every food row.
Christmas Day is optional
Skip logging it or log one rough number: here's what each does to your averages. And why the scale is up on the 26th: a USDA-built Christmas dinner weighs 1.3 kg, glycogen is stored with three to four times its weight in water, and an NIH study measured what the whole season adds.
What connecting your food log to an AI actually hands over
Before you connect a food tracker to Claude or ChatGPT, ask four questions: what the grant reaches, how long it lasts, where the data goes, and how you end it. We asked them of our own connector. Two of the answers were wrong, and one of them was a feature we didn't have.
The Thanksgiving math problem
We built one Thanksgiving plate twice from USDA's own numbers: same dishes, same portions, 1,222 calories or 1,641 depending on four choices, two of which you can't see from your chair. Why the honest way to log the day is to take the high number and stop.
Bad food data wrecks a cut — but not the way you think
The folk theory is that calorie databases are full of strangers' guesses. We counted ours: the crowd is 33 rows out of 1,941,073, and every impossible row we can find sits in the tier labelled authoritative. The expensive mistake isn't a wrong number. It's the right number for the wrong food.
Compounded GLP-1s: what changed, and what's worth writing down
The legal ground under compounded semaglutide and tirzepatide moved between 2024 and now, and a lot of what's written about it online is out of date. Here's the timeline with FDA's own documents attached — and the one part of this that's actually a tracking problem.
The trust problem isn't the crowd
We ran the audit on our own catalog. Search one sandwich and you get 163 answers spanning 157 to 306 calories per 100 g — and most of that spread is real. The rows that are actually wrong come from the sources everyone treats as authoritative.
What the first month on a GLP-1 actually looks like
Week by week, from the trial data and the medication labels: what changes, what's normal, what to track, and what belongs in front of your prescriber rather than a search bar.
A food can't be saltier than salt
About 400 rows in our catalog carried more sodium per gram than pure table salt. Where those numbers came from, the physical rule we now enforce at every write, and the part that isn't finished.
Why search took two seconds
Every food search scanned two million rows and took about the same two seconds, whatever you typed. The fix was to stop scanning for an answer that almost never lives there — with a test proving the results don't change.
Your body keeps a record too
The Body log shipped in 1.9: bathroom breaks, how you feel, supplements and your cycle, one tap each, sitting next to your meals. Built for people whose medication changes more than their appetite.
The numbers you can't read don't count
Font sizes, truncated names, and six 'greek yogurt' rows on six different serving bases. What a real-app design audit found — and the readability release it produced.
Shot day eating: what your own logs can teach you
On GLP-1 dose days, appetite often runs lighter. That's the medicine working — and it makes protein-first eating matter more, not less. How Macroline reframes Today on shot days, and why it stays passive.
Your week: a recap designed to welcome you back
Most weekly summaries are report cards. Ours is built for the person who had a rough week — with the headline that's actually true, and no guilt in the copy.
A food database that grows itself (honestly)
When someone hand-logs a Capital Grille steak because we didn't have it, that's a signal. Inside the pipeline that turns user gaps into researched, verified catalog entries — with the tier labels to match.
Your real burn rate is already in your logs
Every tracker guesses your calorie burn from a formula. We stopped guessing. The coach reads what you actually ate against what the scale actually did — and tunes your target from that.
The most accurate tracker is the one you open tomorrow
Perfect data you quit collecting loses to decent data you keep collecting. On sustainable tracking, the myth of the flawless log, and why we build for the long game.
The two-tap breakfast: favorites, saved meals, and real units
You eat the same breakfast four days a week. Logging it should cost two taps, not a search. A practical tour of the fastest paths through Macroline.
We now log the searches that find nothing
Every failed scan and empty search in Macroline becomes a signal. Here's how user misses now decide what we research next — and why five people beat one person retrying five times.
Your food database should correct itself
Food data drifts — formulas change, labels update, community sources get fixed. Inside the revalidation system we built so stale numbers don't quietly become lies.
Local restaurants are nutrition dead zones. Here's our fix.
Chains publish nutrition. Your neighborhood diner never will. How Macroline turns one person's careful estimate into a shared catalog entry — without pretending it's lab data.
A Fourth of July plate, honestly logged
Cookout food is estimate territory — nobody weighs a rib at a barbecue. Here's how to log the holiday in 90 seconds and why the day after matters more than the day.
Eating enough on the days you can't
On GLP-1s, some days your appetite just doesn't show up. Here's a protein-first floor for those days — no forcing, no shame, just a shorter list of what matters.
When a barcode comes up empty
A scan miss used to be a dead end. Now it's the start of a lookup chain: identify the product, research the nutrition, cite the source — and the next scan just works.
You don't eat a thousand foods, you eat the same twenty
People quit tracking because they think it means endlessly searching a database. It doesn't. Here's the retention trick: set up your staples once, then re-log them in seconds.
How to track restaurant food you didn't cook
The hardest meal to track is the one someone else made. No label, no scale, no barcode. Here's how to log it honestly instead of giving up on the whole day.
Protein and muscle on a GLP-1: what the scale won't show you
On a GLP-1 the scale moves on its own. What it won't tell you is how much of the loss is muscle. Here's why protein matters more when you're eating less, and how to actually hit it.
What we built — Macroline 1.0 ships today
Today Macroline 1.0 lands on the App Store. Here's what's in the box, what we deliberately left out, who 1.0 is for, and the design principles behind the cuts. Substance, no hype — that's the brand.
No tracking, no theater — the macroline.app stack, in full
Most sites that show you a consent banner are loading third-party cookies before you click anything. macroline.app sets zero non-strictly-necessary cookies. Here's what we actually use, what we deliberately don't, and why the popular "Accept all / Reject" pattern is theater more often than it's a defensible privacy practice.
Calorie targets after bariatric surgery — why your app should never refuse the number
A bariatric beta tester told us the iOS app wouldn't let her set 880 cal/day, even though her surgeon prescribed it. The bug is on us. But it's also a pattern in nearly every tracker on the App Store, and it tells you something about how the industry thinks about you.
What "losing strength on a cut" actually looks like in numbers
Lifters dread a strength drop on a cut, but the actual losses you should expect are small and recoverable. Here's the data on what to anticipate, when to worry, and when the drop is just water + glycogen.
GLP-1 and macro tracking — why precision matters more, not less
GLP-1 medications change appetite, satiety, and food choices. Here's how macro tracking should adapt — and why "good enough" calorie estimates aren't.
What 'verified' nutrition data actually means
Most macro trackers display numbers without telling you where they came from. Here's why that's a problem, and how Macroline shows its work on every food row.
Why we built an MCP server for nutrition data
Macroline runs as a Model Context Protocol server. Here's what that means, why it matters, and how AI-native tools change what a tracker can be.
Eating out on a cut — ordering strategies that don't blow your day
Restaurants are where most cuts go off the rails, but they don't have to. Here are the tactics that work without making you the difficult one at the table.
AI-native vs. AI-bolted-on
The difference isn't a chatbot button. It's whether the product was designed assuming an AI agent would be a primary user. Most "AI features" still aren't. Here's how to tell, and why it matters.
Refeeds vs. diet breaks — which one actually helps
A weekly carb-up day and a 14-day return to maintenance look like the same idea, but the research is clearer than the bro-science. Here's what each does, when each works, and when you don't need either.
GLP-1 plateaus — when the scale stalls and what it means
Most people on Wegovy, Ozempic, or Zepbound hit a plateau at some point. Some are signal. Most are noise. Here's how to tell which is which without panicking.
Building an MCP server — lessons from the trenches
Notes from implementing a production-ready MCP server with OAuth 2.1, dynamic client registration, and a real consumer app on the other end. The places we got it wrong, and how the spec helps and hurts.
Bulking with precision — why "see food diet" is wrong
The old advice was eat everything, gain everything, cut later. You can build muscle on a much smaller surplus than that — and the fat tax of dirty bulks isn't worth the marginal muscle.
The maintenance phase nobody talks about
Most people who lose weight regain it. Not because they fail at the cut — because they treat the period after the cut as "back to normal." Here's how to think about maintenance instead.
How to read a nutrition label like a dietitian
The label gives away more than most people realize — once you know what to look for. A practical walk-through of the lines that matter and the ones that don't.
Designing for AI clients vs. human clients — same data, different shape
A REST API for humans and an MCP server for AI agents look superficially similar. They aren't. The contracts, error semantics, and data shapes that work for one are quietly bad for the other. Here's what the second-class consumer of your API actually needs.
How MCP changes the API economy
The Model Context Protocol isn't just another integration spec — it's a shift in who the primary user of an API actually is. Here's what that means for product builders.
Why protein targets matter more after 40 — especially for women
Sarcopenia starts earlier than most people realize, and the standard 0.8g/kg recommendation is set up to fail you. Here's how to think about protein, age, and lean mass.
When AI is wrong about food (and how to spot it)
AI is good enough at parsing meals into macros that the failure modes are no longer obvious. The wrong number doesn't look wrong. Here's a taxonomy of the specific ways AI gets food data wrong, and what to look for in your own logs.
Your TDEE math is wrong (and that's not your fault)
Mifflin–St Jeor + an activity multiplier is the standard, and for most women in their 30s and beyond it's off by 200–400 calories. The equation wasn't built for you. Here's what it gets wrong, what to use instead, and how to find your number with one week of data.
MCP turned one — what shipped, what didn't
Anthropic announced the Model Context Protocol in late 2024. A year and a quarter in, the protocol has stabilized in surprising ways and stalled in others. Here's what's actually in production, what got built and abandoned, and where we think the next year goes.
When January falls apart — what to do in February
The annual cycle is predictable. Strong start, slip mid-month, full collapse by the end of January, vague guilt through February. Here's how to actually use February to fix what January's structure couldn't.
Tools we wish existed in the macro-tracking category
An honest builder's wishlist for what's still missing in nutrition tracking, from substitution preview to cross-tracker import. Open product brief, not a roadmap.
Why we wrote our own nutrition parser instead of stuffing GPT into it
A vanilla LLM call looks like a free meal parser until you ship it. Here's what breaks (units, portions, brand canonicalization) and why a structured parser with confidence scoring is the right tool.
No posts match this filter — try "All" or a different topic.