tinkrbox.dev

AI for the rest of us

Practical local AI tools for images, audio, language, and creative workflow experiments. These apps run in your browser or on your computer so more of your work stays close to you.

Check Edge Readiness

Web Apps

Local-first Edge AI Lab

Use the tools directly

Each app opens from its card. No account flow, no marketing maze, just the thing you came to try.

Know what needs WebGPU

Cards call out heavier local AI features so you can spot browser and device requirements early.

Keep experimenting

Pose, edit, generate, summarize, and synthesize with small tools built around hands-on tinkering.

Windows App directory

Explore the local AI Windows tools

Installation: Download the zip and extract all in the folder of your choosing. Then simply double click the .exe file to run the app.

Windows: These are unsigned tinkering-project apps, so Windows SmartScreen may warn that the publisher is unknown. That warning does not mean the file is known to be unsafe; it means Windows does not recognize it as a signed, widely downloaded app. The SHA keys are provided for verification.

Virus scanners: EdgeExplorer launches a background service to handle model inference for language and diffusion models. Some virus scanners do not recognize them and may complain or try to quarantine the files. Whitelist them when asked to avoid issues.

Readiness

What Can You Run on Your Device?

Familiar with your hardware? Web App Version

EdgeReady

3MB

EdgeReady for Windows checks your hardware locally and reports an estimated AI readiness tier. It reads CPU, RAM, and GPU information on your computer and does not upload that information.
If you are familiar with your hardware (or once you are), the web app can save your tier and provide guidance for the web apps on this site as well (only if you select "Save Browser Profile").
SHA: 6A73EC148B4D74E182D21AE38C76C3DDD12E261CA172E37839E9B6628910056D

Explore

Language, Image (diffusion) and Text-to-Speech Core Windows App

EdgeExplorer

1.3GB

EdgeExplorer for Windows installs the necessary components to run language, diffusion and text-to-speech (TTS) models. A language and diffusion model will be downloaded on first run of each type based on your hardware. You can also select from a currated list of models depending on your hardward. Background removal and TTS models are part of the initial download. Note, the AI models are ~1-3GB each when downloaded.
SHA: FD0DCBA462EE0ABD34CB54D6793B64DBD848BD0D03D1CF0971EED85B82C35890

Web App directory