Practical overview
When it fits
Use DeVinci for browser-based AI conversations when the device and browser meet its WebGPU requirements, and when on-device inference or optional Internet Computer chat storage fits the task.
Who it is for
- People who want to ask questions in a chatbot while keeping conversations on their device.
- Users who want the option to save and reload chat history through the Internet Computer.
- Developers exploring a JavaScript chatbot built with Web LLM and Internet Computer canisters.
Documented uses
- Have a conversation with an AI model running in the browser.
- Keep chats on the device rather than saving them remotely.
- Choose to save chat history and reload it later through the Internet Computer.
Requirements & limits
The project documentation specifies Chrome or Edge on desktop with WebGPU; it says smartphones and other browsers cannot run the app at present. Logging in is needed for Internet Computer chat storage, but chat storage remains a user choice.
- The selected model runs locally, so whether it works and how quickly it runs depend on the device's hardware.
- If a model does not work, the project suggests trying a smaller one.
- The documentation recommends limiting other open programs and browser tabs to leave more computing resources available.
- The repository identifies JavaScript and the Apache-2.0 license.
Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.
Use DeVinci for browser-based AI chat
Chat with an AI model running locally in DeVinci on a supported desktop browser.
Before you start
- A desktop computer running Chrome or Edge with WebGPU support.
- Sufficient device resources for the selected AI model.
Steps
Open DeVinci at https://x6occ-biaaa-aaaai-acqzq-cai.icp0.io/.
Allow the selected model to load in the browser; it is cached there and runs on your device.
Chat with the model. If a model does not work on your device, try a smaller one.
Expected outcome
You can interact with an AI model running in your browser; chats remain on your device unless you choose to log in and save them on the Internet Computer.
Guide limitations
- The documented support is limited to Chrome and Edge on desktop; smartphones and other browsers cannot run the app.
- Model speed and whether a model runs depend on device hardware. Other programs and browser tabs can reduce available resources.
Start DeVinci with a local Internet Computer replica
Prepare the project and start a local Internet Computer replica.
Before you start
- The DeVinci project repository.
- npm and the Internet Computer SDK tools, including dfx.
- Vessel, which the project identifies as a dependency.
Steps
Install project dependencies with the command below.
npm installInstall Vessel by following its project instructions.
Start the local replica with the command below.
npm run devTo start with an erased local replica state instead, run the command below rather than the command below.
npm run erase-replicanpm run dev
Expected outcome
A local Internet Computer replica is started; the standard command retains canister state from previous sessions, while the alternative starts with state erased.
Guide limitations
- These steps start the replica only; deployment of canisters is a separate documented stage.
- The optional local Vite UI is noted as having had backend communication issues in some setups.
Sources & review
Source record details
- Source availability
- Active
- Sources checked
- 15 October 2026
- Original website title
- patnorris/DecentralizedAIonIC · Official homepage title · retained in its source language