Feature Requests

Overview It would be cool to be able to “lock/freeze” assistant settings within the VAPI API dashboard, sucht that there are no unintended adjustments to critical settings once an assistant is in production, ensuring stability and consistency in its operation. Feature Details Description: Introduce an option to “lock” settings for assistants designated as being in production. Once locked, settings would be read-only, requiring an authorized action to unlock them for further adjustments. User Access: Only users with a specified authorization level (e.g., Admin) should have the ability to lock or unlock settings. Confirmation Prompt: To activate the lock, require a confirmation prompt and potentially a reason for logging purposes. Version Tracking: Maintain a record of locking/unlocking events to track any setting changes for auditing purposes. region).

Organizational Tools for Assistant Management Folder Structure: Allow users to group assistants into folders or categories (e.g., "Production," "Testing," "Archived") to better manage large numbers of assistants. Tagging System: Enable users to tag assistants with custom labels for enhanced categorization (e.g., by project, department, or

add timestamp to custom-llm request messages payload current messages array doesn't have timestamp, we want to it to have timestamp per each message so that we are able to insert system message to the right position this is due to the fact that messages from vapi are dynamic, the message received from time 0 can be different from at time 1, in an non-incremental manner, meaning the old message can be override or completely tweaked by new messages. having a timestame would be extremely beneficial for us to manage this dynamism, especially when we insert system message to the right spot of the message history

When twillo voicemail detection updated like below, send a server event: [LOG] Twilio detected "machine_end_other" in 5000ms [LOG] Twilio detected "unknown" in 5000ms

Allow us to call on a custom Assistant using Assistant API instead of only the base model GPT. This will allow us to leverage RAG and function calling and especially Thread ID, so we can have consistency across multiple conversations with the same customer. Alternatively, introduce an equivalent to Thread ID so we can achieve the same consistently natively with VAPI. The use case for clarity is the ability to save the ThreadID to the customer contact in the CRM and then recall it on every subsequent conversation.

It would be helpful to be able to spin up organizations on behalf of my customers where every customer is an Organization in Vapi (basically subaccounts).

Be able to edit more transcriber Settings via the dashboard rather than the API, to enable those who cannot do this via the API. For example end pointing settings for deepgram, transcription is key to the success of how the AI can interact so be great to be able to configure more via the dashboard.

Managing Orgs through API is a key capability to accurately manage and automate onboarded customers. Thanks for considering this as soon as possible.

Enable the agent to navigate any IVR using DTMF tones.

We would like to ability for our team to listen to live calls and transfer ourselves.