# Our AI Features {% hint style="info" %} **Detection Speeds:** Taking advantage of YOLO we can process and detect images at under 80ms speeds in real-time on almost any system! ***Join Our discord:***[ ](https://discord.gg/dFg8smncrR)[***https://discord.gg/dFg8smncrR***](https://discord.gg/dFg8smncrR) {% endhint %} ## Tailor AI Detection We focus on building AI models that differ for every category. Running on [YOLOv8](https://docs.ultralytics.com/) small dataset models, we're able to also integration our own algorithms in between taking advantage of the [RTX 40 series](https://www.nvidia.com/en-us/geforce/graphics-cards/40-series/), and its CUDA cores. We're able to train models our models under larger epochs and configurations to prevent overfitting and much more. With these ingredients, we're able to develop one of the fastest real-time object detection applications for any use-case. That just so happens to have "[Vision Bot](#vision-bot-axis-adjustment)", something you can read on as below. (With options in future versions to run nanite-large models for better accuracy and smaller-beefier devices.)

Vision Assist Real-Time Detection (No Target Locking) (Small Pre-Trained Dataset) Source: Finals

## Vision Bot Axis Adjustment With [Vision Assist](https://visionassist.gitbook.io/vision-assist-v.0.01/vision-assist-v0.01-documentation) v0.01 being focused on the aforementioned aspects, we included a featured marked [Vision Bot](#vision-bot-axis-adjustment), which uses the previously detected object and if enabled, adjusts the values of the axis without manipulating any game files or such, essentially providing the user faster and better aim. This specific feature will always under-go updates and changes to make sure we can provide the optimal solution to adjusting the axis in real-time and preventing over-adjustment (Which is possible when [sensitivity ](https://visionassist.gitbook.io/vision-assist-v.0.01/fundamentals)is set too high)