Weāre excited to announce the release of the Vapi Assistant AI Plugin, designed to help you integrate cutting-edge AI assistant functionality into your applications with ease.
Whether youāre looking to add voice interactions, conversational AI, or automated assistant features, Vapi Assistant AI offers the tools you need for an intelligent, interactive experience.
Top Functionalities:
Assistant Management: Easily create, edit, and delete AI assistants, giving you the flexibility to tailor assistants to specific use cases.
Speech Recognition: Implement voice recognition featuresāstart and stop AI calls, and manage speech events for natural and efficient voice interactions.
Conversation Handling: Keep conversations flowing with real-time updates, ensuring the AI provides smooth, relevant dialogue.
Error Handling: Automatic error detection and management tools to keep your assistant running reliably and smoothly.
Perfect for anyone looking to add voice interaction, automated assistants, and conversational AI features to their apps, the Vapi Assistant AI Plugin brings cutting-edge AI integration to your fingertips.
Hello. The plugin looks good. I am having difficulty with Create Assistant, though. I have set the voice provider to 11labs (and I have set up my ElevenLabs provider credentials in Vapi successfully). When I attempt to create assistant, I get an error stating that I have to declare the ElevenLabs model, but there is no place in the element to enter such information, e.g. eleven_multilingual_v2. Attaching the error message. Appreciate your help on this. Thanks!
Digging further, I am able to create an Assistant with the default values. But now I am thinking that the call is hardwired to use only Azure voices, and of these, only three voices are available. Is that correct? A huge limitation if true, and odd that the plugin would be hobbled in this way. I may be misunderstanding! Please let me know, and thanks.
*See the default voice_provider and voicid fields in the attached screenshot, and the documentation that shows available voice options: andrew, brian, emma. When I try to use an 11labs as voice_provider, and provide a voice ID, the plugin errors out with the error I screenshotted in the last message. Doesnāt seem to be able to handle anything other than azure for voice, and for that, looks like 3 voices.
From your messages, we understand that you initially had difficulty creating an assistant and later identified the cause with the voice provider/voidID fields, is our understanding correct?
We have reviewed your request to examine how the voices are functioning with the Vapi plugin. It appears that some voices from different providers are currently not working. I have forwarded this concern to our development team for further investigation, and we will update you with more information as it becomes available.
If you have any other questions in the meantime, please let us know.
In general, the Create Assistant action is quite limited, should point out. There are numerous parameters that one might want to set, as described in vapiās publicly available API documentation. Even with languages properly usable, that action will be quite limited.
I would encourage you to expand the set of parameters available to make the Create Assistant action worthwhile (tools, background noise, noise attenuation, sign off messages, uploaded context files, etc). Your plugin action is the main differentiator between the ZQ and other VAPI plugins, which are mostly designed to embed the VAPI runtime for an assistant created on VAPI.
A well-implemented VAPI create assistant action will allow integration with our appsā workflows, specialized agent functionality, end user creation capabilities, etc.
We have forwarded it to the development team for future consideration. Indeed the plugin is new, and we always strive to improve our products and plugins. We will keep you in the loop regarding the concerns you had regarding the āvoice provider/voiceIDā fields and the other observations you mentioned in your last message. We will inform you of any updates or decisions from the dev team on them.
Thanks again. If you have any other questions in the meantime, please let us know.
We are writing to let you know about the concern you previously reported with the Vapi plugin, specifically regarding the voice provider and voiceID. In the latest updateāversion 1.5.0āwe have added some improvements that allow you to select all the options available for Vapi and choose their respective voiceIDs.
Could you please update the plugin in your app and test them out? For more information, please check the field docs or documentation, as we have added detailed information on the possible options there.
Also, could you please provide more information on the other improvement you mentioned? If you would like to see a specific option added, we also encourage you to complete the Sponsoring Plugin Features Form.
We will then review it and provide an estimate of the hours required and the associated cost.
Thank you for making the change. I believe the ability to specify a voice provider and voice ID is minimum requirement to make the Create Assistant functionality useful. I am happy to have helped your team improve your product.
That said, there are in fact numerous relevant settings one may make use of when creating a VAPI assistant. In order to implement those I believe make it possible to fully integrate this action in a meaningful way, I went ahead and implemented my own private plugin and Create Assistant api call (Update Assistant as well). So I am in reasonably good shape at this point, with a great deal of expressive range possible in the Assistants I am able to create from within my Bubble apps.
With respect to specific features, well, I could readily detail some important functionality that would, I am sure, make the ZQ plugin stand out relative to other, simpler implementations. But I donāt think it makes sense for me to both provide those market-winning specifications and also pay ZQ for the honor of doing so. Should be the other way around!
So, for now, please accept my thanks for your attention to the matter at hand. Will look forward to more good experiences with Zeroqode and the Zeroqode team.
@ZeroqodeSupport I will mention one thing that would be a very small update, but which would be greatly empowering. The Make Call API provides the ability to create ātransientā Assistants. This means they exist only for the life of the call. This can be very useful in a wide array of use cases, because such an Assistant can be given highly specific, dynamic, and transient context for use in the call (example, current traffic conditions, current specials, etc.)
Otherwise, there are a lot of weird and unnecessary contortion to facilitate transient data for the single, primary Assistant.
To implement this, you would update your Make Call action to provide a field for the transient Assistantās definition. To use this, the user would enter the keyword āassistantā instead of a static Assistant ID, and then paste in the JSON for the definition of that Assistant.
I wouldnāt include the few fields youāre currently using for Create Assistant, as that is too limiting. People using transient assistants will know what JSON to include in that field.
Thank you for your thoughtful feedback and for sharing your insights.
Weāre glad to hear that the voice provider improvement has been useful, and we appreciate your support in helping us refine the plugin. Itās also great to hear about the custom solutions youāve developed.
Regarding your question on āThe Make Call API provides the ability to create ātransientā Assistants,ā we have forwarded this to our development team for further analysis to check the possibility of integrating it. We will get back to you with any decisions or updates on this matter.
Thanks again, and if you have any questions in the meantime, please let us know.
Thanks for your reply. Transient Assistants with variable, dynamically supplied context take functionality to a new level. Best to you for the week ahead!
ā Ron
Definitely, we always strive to improve our plugin and ensure everything is working smoothly. The improvement with transient Assistant was reviewed by our developers, and we have decided to include it in the next plugin update as it is indeed a good addition. We will let you know when it is available.
If youāre enjoying our plugins and have some spare time, weād love for you to rate them. Simply go to the Plugins tab in the Bubble editor, search for the plugin by name, and award it the stars you think it deserves.
Furthermore, if you would like to share your complete experience with Zeroqode, we would be grateful if you could leave your feedback at: Review.
We wanted to follow up on our recent discussion where you asked about the Transient Assistants. Great news in the latest plugin versionā1.6.0ā we have added a new field, āTransient Assistant JSONā in the element where you can set the JSON of the assistant object. This will be applied when the Assistant ID field is empty. You can test it on our demo page and find more information here: Web SDK ā Vapi
Assistant Management: Easily create, edit, and delete AI assistants, giving you the flexibility to tailor assistants to specific use cases.
Speech Recognition: Implement voice recognition featuresāstart and stop AI calls, and manage speech events for natural and efficient voice interactions.
Conversation Handling: Keep conversations flowing with real-time updates, ensuring the AI provides smooth, relevant dialogue.
Error Handling: Automatic error detection and management tools to keep your assistant running reliably and smoothly.
Should be the other way around!
