Workout progression that grades itself: Villain Arc tells you what to lift next, then checks whether it was right
Villain Arc suggests your next weight from your own training history, then watches your later workouts to find out whether it was right. All on your phone.
# Workout progression that grades itself: Villain Arc tells you what to lift next, then checks whether it was right Villain Arc suggests your next weight from your own training history, then watches your later workouts to find out whether it was right. All on your phone. By Fernando Caudillo Tafoya · FCT Technologies · September 2, 2026 import Callout from "../../components/content/Callout.astro"; import BeforeAfter from "../../components/content/BeforeAfter.astro"; import Diagram from "../../components/content/Diagram.astro"; import FeatureGrid from "../../components/content/FeatureGrid.astro"; You rack the bar on your last set, and the app already knows what you should be lifting next week. That part is not rare. Plenty of trackers will nudge a number up for you. The part almost nothing does is what happens after. Villain Arc goes back, looks at how those next workouts went, and decides whether its own advice was any good. ## Most apps remember. This one is accountable. <BeforeAfter before={{ label: "A workout tracker", title: "It records what you did", items: [ "You type in the numbers, it keeps them", "Progress is a chart you have to interpret", "Any suggestion it makes is never mentioned again", "Whether it was good advice is your problem", ], }} after={{ label: "Villain Arc", title: "It coaches, then checks itself", items: [ "It proposes your next weight from your own history", "You accept, reject, or decide later", "It watches the next sessions to see how that went", "It tells you when its own call was wrong", ], }} /> Add five pounds to a lift and there are only a few ways it can go. You hit it clean. You grind and miss reps. It turns out you had more in the tank than the app thought. Villain Arc names all three, out loud, about its own suggestion. <Diagram steps={[ { label: "You finish a workout", detail: "Your sets, against what the plan asked for", kind: "input" }, { label: "It proposes a change", detail: "More weight, different reps, a different set count", kind: "step" }, { label: "You decide", detail: "Accept, reject, or come back to it before the next session", kind: "gate" }, { label: "It grades itself", detail: "Good, too aggressive, too easy, or not enough evidence yet", kind: "output" }, ]} /> ## The interesting part is when it refuses to answer Anyone can write software that always has an opinion. The judgment is in knowing when you do not have enough to go on. <FeatureGrid items={[ { icon: "scale", title: "It waits for real evidence", body: "Some calls settle after a single workout. Others need two or three before the app will commit, because one good day is not proof." }, { icon: "circle-slash", title: "It leaves questions open", body: "If a later session does not tell it what it needed to know, the suggestion stays unresolved. It is never quietly filed as a win." }, { icon: "brain", title: "Rules first, AI second", body: "The app reasons it out directly wherever it can. AI is the fallback for the cases the rules cannot read, and an answer it is not confident in gets thrown away rather than shown to you." }, { icon: "hand", title: "Nothing changes behind your back", body: "Undecided suggestions come back before your next workout, and logging waits until you have cleared them. Accept all or skip all in one tap. The app never edits your plan for you." }, ]} /> An app that says *I do not know yet* about its own advice is telling you something true. It is also a harder thing to build than one that always sounds certain. ## Why it runs on your phone The gym is the one place your phone has no signal. The basement squat rack, the concrete-walled weight room, the corner of the floor where the bars are. So the intelligence runs on the device. Apple's on-device models, your training history staying on your phone, no round trip to anybody's server. That is a privacy answer, but it is not why the decision was made. A coach who needs a bar of signal to tell you what to lift is not much of a coach. <Callout variant="takeaway" title="The idea worth stealing"> Any software that gives advice can also check whether the advice worked. Almost none of it does, because that is the part where you find out you were wrong. It is also the only way the advice ever gets better. </Callout> ## Where it stands Villain Arc shipped to the App Store in February 2026 and ran there through version 1.3. It has since been rebuilt onto FCT's own account and sync platform, the same layer running underneath every app we make, and the listing is withdrawn while that version is finished. There is nothing to download today, and we would rather say that plainly than give you a date we have not earned. Everything described here is built and running. When it returns, the shape is the one it already had: the lifting and the logging are free, and the coaching tier is the subscription.You rack the bar on your last set, and the app already knows what you should be lifting next week.
That part is not rare. Plenty of trackers will nudge a number up for you. The part almost nothing does is what happens after. Villain Arc goes back, looks at how those next workouts went, and decides whether its own advice was any good.
Most apps remember. This one is accountable.
A workout tracker
It records what you did
- You type in the numbers, it keeps them
- Progress is a chart you have to interpret
- Any suggestion it makes is never mentioned again
- Whether it was good advice is your problem
Villain Arc
It coaches, then checks itself
- It proposes your next weight from your own history
- You accept, reject, or decide later
- It watches the next sessions to see how that went
- It tells you when its own call was wrong
Add five pounds to a lift and there are only a few ways it can go. You hit it clean. You grind and miss reps. It turns out you had more in the tank than the app thought. Villain Arc names all three, out loud, about its own suggestion.
The interesting part is when it refuses to answer
Anyone can write software that always has an opinion. The judgment is in knowing when you do not have enough to go on.
It waits for real evidence
Some calls settle after a single workout. Others need two or three before the app will commit, because one good day is not proof.
It leaves questions open
If a later session does not tell it what it needed to know, the suggestion stays unresolved. It is never quietly filed as a win.
Rules first, AI second
The app reasons it out directly wherever it can. AI is the fallback for the cases the rules cannot read, and an answer it is not confident in gets thrown away rather than shown to you.
Nothing changes behind your back
Undecided suggestions come back before your next workout, and logging waits until you have cleared them. Accept all or skip all in one tap. The app never edits your plan for you.
An app that says I do not know yet about its own advice is telling you something true. It is also a harder thing to build than one that always sounds certain.
Why it runs on your phone
The gym is the one place your phone has no signal. The basement squat rack, the concrete-walled weight room, the corner of the floor where the bars are.
So the intelligence runs on the device. Apple’s on-device models, your training history staying on your phone, no round trip to anybody’s server. That is a privacy answer, but it is not why the decision was made. A coach who needs a bar of signal to tell you what to lift is not much of a coach.
Where it stands
Villain Arc shipped to the App Store in February 2026 and ran there through version 1.3. It has since been rebuilt onto FCT’s own account and sync platform, the same layer running underneath every app we make, and the listing is withdrawn while that version is finished. There is nothing to download today, and we would rather say that plainly than give you a date we have not earned.
Everything described here is built and running. When it returns, the shape is the one it already had: the lifting and the logging are free, and the coaching tier is the subscription.
Frequently asked questions
What does Villain Arc suggest after a workout?
Changes to your plan for next time, based on the sets you just finished and your recent history on that exercise: more weight, different reps, a different set count. You accept, reject or defer each one. Accepting updates your plan. Rejecting leaves it alone.
How does the app know whether a suggestion worked?
It watches your later workouts on that exercise and grades its own advice: good, too aggressive, too easy, or not enough evidence yet. Some calls settle after one workout. Others wait for two or more before the app commits, and if a workout does not give it what it needs, the suggestion stays open instead of being written off as a success.
Does it use AI, and does my training data leave my phone?
It runs the rules first and only reaches for AI when the rules cannot tell. The AI runs on your phone using Apple's on-device models, so your training history stays on your device, and it works with no signal at all. An AI answer it is not confident in is thrown away rather than shown.
What happens if I ignore the suggestions?
Anything you leave undecided comes back before your next workout on that plan, and logging waits until you have cleared it. You can accept all or skip all in one tap. The app will not quietly apply a change you never agreed to.
Can I download Villain Arc right now?
No. Villain Arc shipped to the App Store in February 2026 and ran there through version 1.3. It has been rebuilt onto FCT's own account and sync platform, and the listing is withdrawn while that version is finished, so there is nothing to download today and no date we have earned yet.