Compare two periods side by side

New mesocycle starting. You want to know how the cycle you just finished compares to the one before it — same lifts, same routine, two different blocks. The agent pulls both periods and shows them as a diff so you can see what changed and what didn't.

What the agent needs to hear

Behind your sentence, the agent works backwards to three pieces: the two periods, the dimensions to compare, and which side reads as period A versus period B. Periods are the only piece you have to name — named ranges, calendar units, or explicit dates all work.

Dimensions default to the standard set — training volume, sessions, PRs, adherence. Direction matters because deltas have a sign: « this quarter vs last quarter » means current minus previous, so positive means you went up. Flip the order if you want the other read.

Naming periods cleanly

Calendar units are the cleanest names. « April vs May », « Q1 vs Q2 », « this month vs last month » — the agent resolves them against the calendar. « This quarter » is current Q-to-date; « last quarter » is the previous full Q, edge to edge.

Training markers work too. « The bulk vs the cut », « this mesocycle vs the previous one », « before the deload vs after » — anything you’ve named in your routine history, the agent can bound. Explicit dates are the fallback. When both periods match in length, the agent reports raw deltas. When they don’t — a 12-week block against a 4-week one — it surfaces per-week normalizations alongside the absolutes so the comparison stays fair.

What the comparison actually answers

A diff answers questions a single-period read can’t. « Is the new routine working better than the old one? » — same lifts in both blocks, the comparison shows whether volume and PR counts moved under the new programming. « Did the cut cost me strength? » — bulk against cut tells you whether the heavy work held up under deficit, or quietly bled off.

« Was last cycle better, and why? » is the broadest read. The diff surfaces which dimensions improved and which regressed — volume up, PRs flat, adherence down — and the shape of those deltas tells you what to keep and what to redesign. The question you bring shapes which dimensions matter; phrase it specifically and the agent narrows the diff.

When the agent gets it wrong

Periods that overlap are the first failure mode. « Q1 vs the first half of the year » double-counts January through March on the second side — name disjoint ranges (« Q1 vs Q2 ») and the agent re-pulls.

Dimensions that don’t fit the question are the second. You asked whether the new routine is improving strength, but the card leans on session count and volume — narrow it: « just PRs and heaviest sets on the main lifts ». The third is direction. Confirm which period is the baseline: a « -5 % volume » delta means current is below previous if A is the older period, but the opposite if you flipped the order. Say it explicitly when it matters.

What makes the comparison worth reading

Three things decide whether this comparison tells you something useful: the periods are comparable in length and shape (a 12-week mesocycle against a 4-week deload won’t give you a fair read), the dimensions match the question (asking « is the new routine better » means looking at volume and PRs on the same lifts, not raw session count), and deltas are read in the right direction (improvement is positive on volume and PRs, but negative on body fat or recovery time). Read the diff, decide what carried over and what didn’t, and let the answer shape the next cycle’s design.