Workout tracker app for Claude and ChatGPT: log workouts by chat

AIm is a free workout tracker app with an unusual interface: the chat assistant you already use. You tell Claude or ChatGPT what you lifted in one sentence, the app keeps the structured log, the charts and the records, and every future chat can read your history back.

Updated: 2026-08-14

Logging by sentence instead of by form

This is the part that decides whether a log survives past week three. A traditional tracker asks you to find the exercise in a list, add a set, type a number, repeat. It is accurate and it is friction, and friction between sets is what kills logging habits.

Describing the session costs one message:

The assistant does the structuring: exercise names, sets, reps, weights, the note that the last set was hard. You never open a form. If you got something wrong, you say so in words rather than hunting for the row to edit.

The tradeoff is honest: typing a sentence is slower than tapping a familiar app if you are logging mid workout, set by set. It is much faster if you log once at the end, which many people do anyway.

Your history follows you into every new chat

Claude stores the session: exercises, sets, weights, body weight. A week later, in a brand new chat:

Claude reads your history through the connector and answers with your real numbers. Mistakes are fixed the same way:

Charts, records and a program in a real app

Everything the assistant logs shows up in a mobile web app: a muscle map of your weekly volume, progression charts per exercise, personal records, body metrics, your current program and progress toward the goal you set with the coach. You talk to the assistant, the app keeps the picture. And the data stays yours: export everything as Excel or JSON whenever you want.

That split matters more than it sounds. Chat is a good interface for writing and for questions, and a bad one for looking at a trend across four months. A screen is the opposite. Putting the assistant in front of a real app gives you both, instead of asking one of them to be something it is not.

AIm works in English, Portuguese, Russian, Spanish and French.

How the connection works

Claude supports connectors built on MCP, an open protocol that lets the assistant use external tools. AIm is a workout storage connector: it gives Claude tools to write and read your training log, kept in private storage that only your link can reach.

After connecting, you just talk. Claude logs each session through the connector, and every future chat can read the full history back.

Why the chat forgets your training

Every chat is a separate conversation. When you close it, the sets, weights and pain notes you described stay behind in that chat. The next conversation starts blank.

Most people training seriously end up with two things open. A workout tracker app, where you tap in the exercise, the set and the weight. And a chat assistant, where you ask what to do next and why your bench has stalled.

Neither knows the other exists. The tracker holds a clean history no assistant can read. The assistant gives advice built on whatever you remembered to type into that particular conversation. You end up as the integration layer, copying numbers between two apps, which is exactly the job software should be doing.

• Pasting history into each chat. Works for a week, then the wall of text grows, eats the context window and drifts out of date.
• Keeping a spreadsheet. Now you maintain two systems by hand and the assistant still cannot write to it.
• Projects with attached files. Better, but the file does not update itself after every workout, and there are no charts or records.
• The built-in memory feature. Both ChatGPT and Claude now remember things about you across chats, and people reasonably assume this solves it. It does not, because it is designed for preferences rather than records. It will happily remember that you train four times a week. It will not reliably hold sixty sessions of exercise, set, rep and weight, and you cannot chart what it stores.

There are two ways out. Either your tracker learns to expose its data to an assistant, which is what the community bridge servers in the connector guide do, or your assistant gets a place to write. This page is about the second, because it takes about a minute to set up and the assistant can log new sessions, not just discuss old ones.

What to look for in a workout tracker app that works with AI

If you are comparing options rather than taking a recommendation, these are the questions that actually separate them.

• Can the assistant write, or only read? A read-only connection can discuss your history. It cannot log today's session, which is most of the value.
• Is the data structured? Free text notes cannot drive progression. Exercise, set, rep and weight need to be separate fields for anything to compute an estimated one rep max or a weekly volume.
• Does it work with the assistant you already pay for? A tracker with its own built in AI is a second subscription and a second place your history lives.
• Can you get your data out? Export should be one click or one sentence, not a support request.