Using WebLinx models for chat-based web task assistant.
To use or develop the extension locally, please refer to its documentation. To use or develop the backend locally, please refer to its documentation.
Our main purpose is to have a very intuitive frontend, and remove as much friction for the user. Hence, we chose to develop a Google Extension.
On a high level, a typical workflow would look like the following:
For communication purposes, the extension and backend currently communicates through dedicated REST API calls. Details can be found here.
The backend server would be responsible of keeping track of the turns, and communicate with the two models. The following diagram is provide for clarity: