Kaltro’s suggestion engine is deterministic and runs entirely on your iPhone. These four write-ups quote the shipping rules: when it adds weight, when it deloads, how much of your history it needs before it says anything, and how it turns your logged sets into an e1RM trend. Every rule is a tunable design choice, shown in the open.

When Kaltro adds weight

The double-progression gate: every working set at the top of its rep range, a consistency count set by your experience, your nutrition phase, any injury you are ramping back from and how your own recent sessions have gone, and an effort ceiling of RPE 8 for most lifters before the weight moves.

Read how Kaltro adds weight

When Kaltro deloads

Trigger-based, never the calendar: rep regression at the same weight, time away on a mark that varies by experience level, or logged effort staying high across consecutive sessions — a session count that grows with your training experience, plus one more session in a calorie deficit.

Read how Kaltro deloads

Minimum evidence before a suggestion

Why Kaltro stays quiet at first: the three-session floor, the per-decision minimums behind it, and two earlier exceptions — a lighter restart after time away, and the lighter target of a Deload phase you set.

Read how much history Kaltro needs

Reading the e1RM trend

How logged sets become an averaged Epley-and-Brzycki trend for 1 to 12 reps, and why to read its direction rather than any single point.

Read how the e1RM trend works

New to any of these terms? The training glossary defines them in plain English, each matched to how the app computes it. Switching apps? Bring your history in from Strong or Hevy and the engine starts reasoning from it right away — or see what Kaltro does with your training history.

See the reasoning on your own lifts

Every suggestion Kaltro shows comes with its reason, built from your own logged sets and one tap away at most. The suggestion itself comes from the rules described here, running entirely on your iPhone.