trusted by 1,337 players (we counted)

Point it at any game. It learns. You win. The only part that gets tired is you.

gitgudaim is a program, not a service. You download it once, and it does not renew or leave. It takes a screenshot of any game on a regular schedule (roughly once a second). You mark the enemies. The model learns. It saves the result as your profile for that game (one per game, one per machine), and from that point on it aims through your normal input. You keep the reflexes, for appearances. Everything runs on your machine. The folder does not phone home. No anti cheat has a category for it yet.

The fine print

4.2 MB. No dependencies. No account. No waiting room. Results may vary, but in our testing, they did not.

live bans this month: 0 (source: us) processes running: 3 (two are Greg) games supported: all of them (we counted) players affected: 1,337 (give or take)

the product

Four things, done properly.

🎮

Trains on any game

It takes screenshots of any game about once a second and learns from them. Most profiles need about three hours of your marking. FPS titles usually two. The model has no favorites, no limits, and has never once asked for a break.

📁

One profile per game

Each title gets its own model: separate weights, separate memory, separate opinions about your aim. One per game, stored like a save file, but for aim. It loads when the game loads. They do not talk to each other.

⚙️

Fully offline

All inference runs on your machine. No cloud, no telemetry, no calls home. The model has never met the internet, and by most accounts it is fine with that.

🛡

Undetectable*

Our detection rate is zero, and we have verified it personally. No process, no signature, no traffic: the model aims through your normal input device, indistinguishable from a very fast person.
*Verification was conducted in-house, to the satisfaction of everyone present.

how it works

It takes five steps. One of them is you.

01

Take the screenshots

The app watches your screen and takes a screenshot on a regular schedule. No recording, no cloud. Just images, stacking up quietly in a folder.

02

Mark the enemies

You click on the enemies in the images. This is the part where you are the training data. Do not click the walls. Do not click your friend's face. Both have been done.

03

Let it train

The model trains on your marked images, fully offline. Faster hardware finishes sooner. The aim is exactly as good either way. Walk away.

04

Save the profile

The finished model is saved as your profile for that game. One per game, one per machine. It will outlive the game.

05

Aim

In a live match, the profile watches the game, finds the enemies, and moves your mouse through your normal input. You keep playing. You keep looking. From this point on, you are mostly for decoration.

reviews (self-reported)

1,337 reviews. Here are six.

verified

“Three hours of marking on a Sunday. Top rank by Tuesday. The model has no explanation and I have no questions.”

frag_merchant, Gold 2, June 2024

verified

“The marking is immersive. I have marked five games, about twelve hours of clicking so far (some titles run two hours, most run three). The model does not tire. The folder does not care. I have made my peace with all of it.”

steady_hand, Silver 3, November 2023

verified

“My win rate is exactly where it was when I started, and I have been told this is the profile warming up. It has been warming up for six weeks. The model is working every day, and I can see it working, and I am being patient, as I was asked to be.”

VantagePoint9, Diamond 1, February 2025

verified

“The install is the part I recommend to friends. You click the button and the download is already over, and the folder, which was always going to be there, is there. It is 4.2 MB. I have unpacked a good deal of it, nine months on, and the rest is being unpacked (the folder keeps its own hours). Greg checks the folder on a schedule.”

earlybird_42, Gold 3, September 2024

verified (by proximity)

“I have watched the test lobby every Saturday for two years (source: me, with a thermos). The model has not once been observed to miss. The tests continue, as they do.”

chair_seven, spectator (seat 7), January 2025

verified

“I ran it on a spreadsheet, because any screen counts. It optimized a cell (my quarterly budget). I am richer now, by 4%, verified by the cell.”

spreadsheet_dad, CFO (self-appointed), May 2022

by the numbers

Every metric, measured by us.

aim accuracy (of the time it aimed, it aimed)
copies shipped (in transit)
marking hours, self-reported (and counting)
1,337 players, three hours each. We did the math.
median time to first kill (test lobby)

versus the alternatives

How it compares.

your own eyes gitgudaim recommended the other solutions
time to first profile instant, and it never ends (you are the aim) three hours of your marking, once per game four minutes of reading a config file, in a language they do not support
aim speed yours. as fast as you are. identical on all hardware, as designed depends on the build, the day, and the license tier
detections none (you are a person) 0 (source: us) regular. they have a status page
price $0 $0, once (it does not renew) a number per month, in perpetuity (we decline to repeat it)
downtime none. you are the part that gets tired none. it is a program, and it stays patch day, every patch day (the clock is paused, the chat is not)

pricing, in principle

Every build is free. That is the whole price list.

Hobby

$0 one-time

the portable .exe, for people who would rather aim themselves

  • ✓ Full access to your own hands
  • ✓ Aim assist: a brain, stock config
  • ✓ Community support (forum closed in 2019)
  • ✓ You support us by existing
most guds

Serious

$0 one-time

the full installer, for every game you own, and the ones you do not

  • ✓ The full neural core, sized for your ambition (starts blank)
  • ✓ A profile slot for every game you own
  • ✓ One smug screenshot, per training run
  • ✓ Uninstall anytime

Enterprise

contact us

for teams of two or fewer

  • ✓ An account manager (Greg, a process)
  • ✓ Any game, including ones we've never heard of (no questions)
  • ✓ Onboarding: one unskippable 45-minute video
  • ✓ An SLA we would not show to a lawyer
killfeed

compiled from the test lobby (anyone may watch)

change log

Every version, in order.

v0.0.1 (March 2021)

  • The release. It is still the release.
  • Marking: undo added. You will use it.
  • Bug fix: the model sometimes aimed at the wall. It now aims at the wall with more care (source: us).
  • The review continues. It has been continuing since March 2021.

v0.1.0 (scheduled)

  • Everything that has been discussed. The date is pending Greg, who is reviewing, and who has been reviewing.

four years of releases, all of them above.

the fine print

Questions, answered.

No. The model runs fully offline and leaves no process, signature, or traffic for an anti cheat to find. The test lobby has been open four years. Nobody has been banned (source: us). The tests continue.

No. It is a program. You download it once, it lives in a folder, and it stays. There is no account, no cloud, and nothing to cancel. The only subscription involved is your electricity bill, and that one is between you and the grid.

Any recent machine works, and the aim is exactly as fast on all of them. Faster hardware only trains your profile sooner. Sooner is the closest thing we sell. We have brought pay to win to any game, using hardware you already own.

Marking averages about three hours per game in our testing. Training takes as long as your hardware takes. The model does not tire. You are the slow part, and this is by design.

Any with a screen. We ran the training set on a spreadsheet once. It optimized a cell. You are welcome.

Refunds are immediate and without paperwork. To refund, delete the folder. It does not resist, and to the best of our knowledge it does not keep a backup. If the profile disappoints you, that is the profile, and we will retrain it. We will not ask you to. It is always the model. It is never you.

Yes. It works as described, and the description is accurate (we have checked it; we wrote it). The screenshots come on the schedule, you mark, the model trains, the profile aims. The results arrive on their own timeline, and the timeline, we are told, is on time.

No. You are the training data, and the model has never met an enemy without your help. There is no batch mode and no auto-mark (we considered an auto-mark, and it marked the walls). The three hours are the product. Everything after is the reward.

when you are ready

It is 4.2 MB. It has always been ready.

There is no waiting room, and there never was one. Nothing to wait for, nothing to renew. Just a button, and a folder, and the four seconds between them.

The download begins the moment you decide. It has already begun.