ambitious-aqua•2y ago
Voiceflow Agents in Production 🎬 - Voic...
A bunch of updates for you all this week! @everyone
Project Create Update
As mentioned in the large post above we have updated our project create. We are focusing on webchat and our API and have removed our other integrations. To connect to other channels you can use our API directly or with no-code tools like make, zapier, or manychat as user have demonstrated in #📚┃community-tutorials
We have also added a Voice AI project type! This allows you to upload and play audio files. Perfect for creating Voice experiences with Open AI's Voice models, Eleven Labs or others. This is seperate from Webchat. So its mainly meant for if you're building a Voice-only experience with the API.
You can read more on why we did this here: https://changelog.voiceflow.com/en/voiceflow-agents-in-production-KxR8QGMX?role=s_pGLwcAKU18186
A great tutorial on using manychat to connect VF to Instagram here from @emilwiz [Lvl. 1] : https://www.youtube.com/watch?v=L2PETLGCc7Q
Our dev docs with examples on connecting to whatsapp, slack, discord, etc. here: https://developer.voiceflow.com/
Send variables into Voiceflow through the Chat Widget
This update lets you send information stored on your webpage into Voiceflow when the chat widget loads! This can be things like url, email, name, etc. and access them in the Javascript step.
Tutorial here: https://youtu.be/vJ7moFgpoT4
Create Dynamic Steps on Voiceflow with JSON (for Developers)
This update is incredibly powerful - it lets developer use javascript to create any on the canvas. This means that you can do things like take a big array returned from an API with 50 products and turn it into a carousel with 2 steps (javascript + custom action).
Here is a demo that Nico put together using GPT to generate the code: https://www.loom.com/share/06f8d45c795a47528cb8c850bd1a2fd2?sid=811a5bb4-985d-4a99-8f53-f1077ec40ed1
Here is a full overview from Tyler on how it works: https://youtu.be/2V46rOkkprg

1 Reply
provincial-silver•2y ago
Next up:
1. Updates to AI models (GPT 3.5 Upgrade to 16k context, updated token multiplier - we're waiting on GPT4-Turbo to be production ready from Open AI so that is not available yet)
2. Quality of Life update to transcripts page
3. Large update to KB step to improve accuracy