Can I use TaskingAI to provide an already created Assistant to others in the form of a webpage?

[Hapluckyy]

I would like to generate a separate interface, like the chatGPT interface, for an Assistant's Playground to be used by others for testing. How can I do this, and do the developers have plans to accommodate this feature?

I would be very appreciated if you can help me out, TaskingAI is indeed a great tool to create LLM apps. Thanks!

[DynamesC]

Thanks for the suggestion Hapluckyy. Currently, after a user finished configuring an assistant, the assistant can be accessed outside TaskingAI with the following two methods: through the Python SDK or through the REST API. At the moment, we do not support to share your assistant directly with a link as GPTs do. That's because we see ourselves as a developer tool, and we prioritized the features that really facilitates developers' works.

But your suggestion is great, as this will help users to share their cases and may inspire each other. We will implement this, but with relatively lower priority. Please keep an eye on our cloud platform for updates.

BTW, could you tell me the motivation or scenario of why you need this feature. This will help us better understand user needs. Thanks.

[Hapluckyy]

The "Retrieval" created by TaskingAI indeed offers a high level of programmability, allowing developers like us to incorporate many of our own ideas and integrate various RAG algorithms. We greatly appreciate the work done by the TaskingAI team.

However, after creating the LLM app, I currently require a visual interface to provide to others for testing and feedback, in order to evaluate the quality of "Assistants" 's outputs and make improvements to my app. The current mainstream LLM UI interfaces are still based on the OpenAI API. Here is an example that I am currently using:

https://github.com/ChatGPTNextWeb/ChatGPT-Next-Web

Following the creation of "Assistants" using "Retrieval", there have been several challenges in integrating the UI interface due to the difference in API formats between TaskingAI-API and OpenAI-API.

If TaskingAI Team can provide a dedicated user testing interface directly through TaskingAI's management interface, which can be accessed via a separate URL, or alternatively, offer an API endpoint in the OpenAI-API format to invoke Assistant, either of these solutions would help me address the current challenges I'm facing.

I'm also very excited about accessing LLM apps designed by other developers and showcasing my own designed LLM app to promote the TaskingAI tool. TaskingAI is indeed a great project that offers a high degree of programming freedom, making it easy to bring my ideas to life.

[Hapluckyy]

By the way, can I offer you guys a cup of coffee? Are there any ways to express my gratitude? TaskingAI is indeed a Great Work!

[DynamesC]

A management feature currently under development is to invite people to your space and projects. Invited users can access and test your assistants. There will also be permission management for different roles. I think this feature will also meet your requirement?

A great THANK YOU for the kind words! Your recognition is already a great motivation for us. We will launch paid plans for our cloud platforms soon. You may select a paid plan or stick with the free plan based on your needs. Please stay tuned.