See progression on a specific lift

You think your bench has been moving but you're not sure. You want to see the curve — what your top sets have looked like over the past few months, not just the latest number. You ask, and the agent draws the timeline.

What the agent needs to hear

Behind your sentence, the agent works backwards to three pieces: which exercise, the time window you want to see, and what metric answers your question. The exercise is the only piece you have to name. The window defaults to the last three months — long enough to show trend, short enough to be relevant — and the metric defaults to estimated 1RM, the standard “am I getting stronger” answer.

The agent reads your set history for that exercise across the window, computes the metric session by session, and shows you a card with the timeline. Each point is a session; each point’s tooltip shows the set it derived from. No abstraction — every datum is traceable back to a real log.

What you can ask for

There are four common views of progression, and they answer different questions:

• Estimated 1RM trend — the standard. Each session’s heaviest working set becomes a 1RM estimate; the curve shows whether your peak strength is climbing. Default.
• Heaviest set trend — actual top weight per session. Less smooth than the 1RM curve (you can hit the same weight for fewer reps and the estimate drops while the heaviest weight stays flat) but truer for max-effort progression.
• Volume trend — total weight × reps per session. Useful for hypertrophy-focused blocks where you care about accumulating work, not peaking.
• Rep range progression — reps at a fixed weight over time. Ask for « rep progression at 80 on bench » when you’re chasing rep PRs instead of weight PRs.

For time, the default window is three months. « Last month » zooms in to recent sessions; « last year » or « all time » zooms out. « Since the start of this mesocycle » or « since the routine activated » bounds it by training-cycle markers.

What the timeline reveals

The shape of the curve answers the question your gut is already asking. A climb is climbing; a plateau is plateauing; a regression is regression. The agent reads the curve back to you in plain language — « bench est. 1RM has climbed 7 kg over 12 weeks, currently 102 » or « squat has plateaued at around 130 for six weeks ». That phrasing isn’t analysis, it’s description, and the dates on the card let you verify it.

If the trend doesn’t match what you expected (you thought bench was moving but the curve is flat), the card is where to dig in. Tap a point and you see the session it came from — sometimes a single bad session pulls the line down for a stretch, and seeing which one tells you whether the trend is real or noise.

When the agent gets it wrong

Read the card before you accept the trend. If the exercise is wrong (« that’s the cable row, I asked for barbell row »), name it cleanly. If the window is wrong (« only show me the last six weeks »), say so and the agent re-renders.

If the metric doesn’t answer your question (you wanted volume but got 1RM estimates), ask for the right one: « show volume instead, not 1RM ». And if a session looks anomalously low or high and the agent didn’t flag it, you can ask: « what happened on March 12? The line dropped sharply that day ».

What makes the trend worth reading

Three things decide whether this progression view tells the truth: the exercise filter is exact (flat bench is not incline; barbell row is not cable row), the window covers the period your gut is actually asking about (not just the default), and the metric matches the question you’re really asking (1RM for strength, volume for hypertrophy, rep PR for rep progression). The system reads your sets the same way every time; the difference is what you ask to see. Frame the question the way it lives in your head, read the curve, and the next session writes itself.