Skip to content
All posts
Journal9 min read

5/3/1 with an AI coach: what changes when programming isn't manual

Running 5/3/1 in a spreadsheet works until it doesn't. Training max management, supplemental template selection, AMRAP interpretation — the parts of the program that require judgment are also the parts that go wrong most often. Here's what changes when that judgment is automated.

GlenBuilding Nahyeh
5/3/1 with an AI coach: what changes when programming isn't manual

5/3/1 is one of the most durable strength programs ever written. The wave loading across three weeks, the AMRAP top set, the training max system that keeps you from grinding yourself into the floor — it's a genuinely well-designed program that works for a lot of people over a long time.

It also has failure modes. Not in the program itself, but in how people run it. Training maxes that creep too high because the AMRAP went well and the lifter got excited. Supplemental templates chosen based on what someone on Reddit recommended rather than what matches the lifter's actual recovery capacity. Deloads skipped because the schedule got tight. Accessories run at random intensity because there was no real plan for them.

None of this is a knock on 5/3/1. It's just what happens when a program that requires ongoing judgment gets handed to a spreadsheet.

This post is about what changes when that judgment is automated.

What Yuge handles automatically

When you run 5/3/1 through Yuge, you start with a conversation rather than a spreadsheet. You describe your training history, your current numbers, how many days you can train, and any constraints — injuries, equipment, time available. The coach asks follow-up questions: how have you responded to high-frequency training before? Do you recover well from heavy volume on accessories, or do you tend to accumulate fatigue there? What's your main goal — strength expression, building the total, general progression?

From that conversation, it builds the program. But the more interesting part is what it does after that.

Training max management. After every AMRAP set, Yuge reads the reps and RPE and updates the training max accordingly. If you hit 10 reps on your top set of squats at what felt like an 8 RPE, it computes the appropriate adjustment for next cycle. If you hit 5 reps and it felt like a 9.5, it holds the training max where it is. If you've missed the rep targets on multiple consecutive cycles, it resets. You don't have to decide. The math runs automatically, and you can see the reasoning if you want it.

Supplemental template selection. 5/3/1 has a lot of supplemental options — Boring But Big, First Set Last, Building the Monolith, BBS, 5s Pro, Pyramid, and others. Most people pick one based on a forum thread and run it until they stall or get injured. Yuge selects the template based on the context: your training history, how you described your recovery capacity, what your goal is for this cycle. Higher-volume templates like BBB tend to suit lifters with solid recovery who are chasing hypertrophy alongside strength. Lower-volume templates like FSL tend to suit lifters running higher frequency or coming off a reset. The coach knows the tradeoffs and picks accordingly rather than asking you to.

Accessory programming. Accessories in 5/3/1 are often the weakest part of how people run the program. Jim Wendler's original books are deliberate about keeping accessories simple, but in practice most lifters end up doing random upper/lower work with no progression logic. Yuge builds structured accessory blocks tied to the main lift pattern: horizontal pulls alongside press days, single-leg and hip hinge accessories alongside deadlift days, and so on. Loads and rep schemes follow progression logic, not just "same as last time."

The parts of 5/3/1 that trip people up

The interesting question isn't whether Yuge can automate the straightforward parts of the program. The interesting question is whether it handles the parts that require judgment.

When to reset training maxes. The most common mistake in 5/3/1 is running training maxes that are too high. Lifters hit a run of good AMRAP sets and keep bumping the TM. Eventually they're grinding near-maximal singles on their week 3 top sets, recovery tanks, and they have to reset anyway — but now from a deeper hole. The heuristic most experienced 5/3/1 lifters eventually land on is: your top set should rarely feel like a genuine limit. If week 3 stops feeling like a hard training session and starts feeling like a test, the TM is too high.

Yuge applies that heuristic with actual data. It tracks RPE across your top sets week to week. It looks at your rep performance relative to predictions. When the signals suggest the TM is drifting high, it proposes a reset before you've ground yourself into the floor — and explains why, so you're not blindsided. More on how this works in the post on training interventions.

Supplemental template progression. BBB is a good example of where people get into trouble. The volume on BBB is high. At 50% of training max it's manageable. A lot of lifters start bumping the supplemental percentages because the work feels easy early in the cycle. By the time they're deep into a BBB run with supplemental work creeping toward 70% of TM, the cumulative fatigue is significant. Yuge runs supplemental percentages according to what the template actually prescribes, adjusts based on completion signals, and keeps the supplemental work anchored to what the template is designed to do — build volume, not test strength.

Deload timing. The standard 5/3/1 deload week every fourth week works for many people but not everyone. Some lifters need less frequent deloads. Some — particularly those managing high life stress, or running the higher-volume templates — need them more often. Yuge monitors fatigue accumulation signals across the cycle and triggers deloads when the data calls for it rather than on a fixed schedule. If week 4 rolls around and every signal says you're recovered and progressing well, it doesn't force a deload. If week 3 shows clear fatigue accumulation, it brings the deload forward.

A concrete mesocycle

When things go wrong

Missed sessions. Miss a session mid-cycle and Yuge redistributes volume across the remaining sessions rather than trying to cram the missed work in on top of the scheduled work. If you miss enough sessions in a short window, it adjusts the weekly volume expectation downward. The program doesn't pretend the missed work happened.

Failed AMRAP. A week 3 AMRAP that comes in significantly under predicted reps is a signal. One bad session isn't a verdict — the coach notes it and watches the next cycle. Two cycles of underperformance triggers a TM review. Yuge doesn't wait for you to diagnose the problem. It flags the pattern.

Plateau on a lift. Stalled progress on a single lift gets different treatment than global fatigue. If your squat has been grinding while deadlift and press are progressing normally, the coach looks at the squat specifically: supplemental volume, accessory work, form notes you've mentioned in conversation. It might propose swapping a supplemental template on the squat while keeping everything else as-is. Main lifts are treated individually. Stalling across all four lifts at once is a different signal — that usually points to a recovery issue.

Something hurts. If you mention that your lower back is bothering you on deadlifts, the coach swaps the movement, explains what it's substituting and why, and notes the issue for future sessions. It doesn't push you through pain. It also doesn't nuke the whole program because one movement is aggravated. See how Yuge handles training interventions for detail on how the substitution logic works.

How this compares to a spreadsheet or Boostcamp

A spreadsheet runs the math correctly if you set it up right. It does not read your AMRAP performance and adjust your training max based on a trend across multiple cycles. It does not notice that your RPE has been climbing and propose a TM reset before you've overcooked it. It does not change your supplemental template when the volume isn't matching your recovery. You do all of that manually, when and if you notice.

Boostcamp runs the program accurately. The 5/3/1 implementation in Boostcamp correctly applies the training max system and the AMRAP-based TM adjustments. That's genuinely better than most apps. What Boostcamp doesn't do is adapt to your specific training history, respond to fatigue signals mid-cycle, help you choose between supplemental templates based on your context, or change course when something isn't working. The program runs as written until you decide to change it.

The difference with Yuge isn't that it knows a better version of 5/3/1. It's that it's paying attention over time in a way that a spreadsheet and a passive program runner can't. Most lifters who have run 5/3/1 for a few years have developed good instincts for when to reset, when to push, which supplemental template fits their recovery. Yuge applies those instincts systematically, with actual data, from the first cycle.

If you're weighing 5/3/1 against GZCL, we've made the same case there — different methodology, the same place where a spreadsheet runs out of room. And for the within-week version of structured variation, see daily undulating periodization.

If you want to run 5/3/1 and have those calls made for you based on how your training is actually going, Yuge is in early access.

Early access

Want early access?

If you're a serious lifter who wants AI that understands your training methodology, we'd like to hear from you. Gym owners should check out Hoist. Early adopters get first access.

Early access. We'll let you know when it's ready.