Strength training has nine decades of published science behind it. Most apps ignore all of it. We built one that doesn't.
Every line of code in Meso's programming engine maps to a peer-reviewed source. If we can't cite the rule, we don't ship the feature.— The whole pitch, in one sentence
The lifters who keep progressing for years don't run linear programs forever, and they don't chase PRs every week. They cycle through phases — build a base, sharpen the work, peak, recover. The data on this is decades old and unambiguous.
Meso treats block periodization as the default structure of the product, not an advanced toggle. Every program runs through accumulation, intensification, realization, and deload — because that's what the science says works.
A program that prescribes "Monday: 4×5 at 82.5%" without knowing how Monday actually feels is guessing. The lifter who slept four hours and trained fasted needs different load than the one who's fresh and well-fed.
Meso autoregulates every set using RPE and RIR — the same tools elite coaches use — so target weights bend toward your readiness instead of forcing it. The program adapts to you. Not the other way around.
Generative models hallucinate. Yours, ours, theirs. That's a feature you accept when you're brainstorming and a bug you cannot accept when you're loading a barbell. So the model doesn't get to decide much.
Inside Meso, an LLM only picks exercises from a pre-validated pool that's already been filtered for your equipment, injuries, fatigue, and program structure. Everything else — sets, reps, weights, deloads, progression — is deterministic code enforcing rules from the literature. The model can't invent a new exercise. It can't override a contraindication. It can't skip your deload.
"More sets = more gains" is the most common bad advice in lifting, and it's why beginners burn out at month four. Volume that's too low gets you nothing. Volume that's too high gets you nothing — and an injury.
Meso uses Mike Israetel's MV / MEV / MAV / MRV framework, recalibrated per muscle group every block, so the dose tracks your actual recovery rather than someone else's spreadsheet.
You'll travel. The squat rack will be taken. Your shoulder will tweak in week three. A real coach handles those without restarting your program. So does Meso.
Mid-session swaps are ranked alternatives, not random replacements. Travel Mode re-plans the whole week for the equipment you actually have. Injury flags filter the exercise pool before the model picks. The plan adapts in real time without losing the thread of your progression.
You don't have to take our word for any of this. The full methodology — every rule, every threshold, every progression scheme — is published with citations and DOI links. If a paper we cite is wrong, you can find that out. If a rule we apply is unsupported, you can challenge it.
Most apps treat their programming as a black box. Ours is a glass one. Read the methodology before you give us money — that's the whole point.
Meso started from a gap visible in every fitness app on the market: they could log a workout. They could not design one. The closest thing to a real program builder was a chatbot, and the chatbot did not know what a deload was.
We reviewed the strength science literature, built the engine, and pressure-tested it on ourselves. The result is what you see: not a chat interface, not a spreadsheet, not a generic workout app. A periodization engine, in your pocket, with receipts.
We're based in Copenhagen. We write our own checks, take no advertising, and answer our own support email. If you want to argue with us about the methodology, we'd love that — methodology@meso.app goes straight to the people who built it.
Read the methodology, check our citations, then decide.