This June, we focused on shipping production-grade tools that give you more power and control over your agents.
Here's everything we shipped this month:
Major Feature Launch
Workflows Are Now Production-Ready
Workflows are out of beta and ready for production. We’ve hardened the infrastructure and added the configuration depth needed to build and deploy sophisticated, multi-step voice agents with confidence.
What you can now do:
- Dynamic, Per-Call Behavior: Use workflowOverrides to inject custom logic, variables, and tools into a workflow for each unique call.
- Faster Development: Use the new canvas-level bulk editing to select and modify multiple nodes at once, speeding up complex flow design.
- Real-Time Debugging: See exactly how your conversations are progressing with live transcripts and animated node traversal directly in the workflow canvas.
- Orchestrate Complex Flows: Reliably route calls, swap models mid-conversation, and integrate external data for any use case, from advanced customer support to detailed lead qualification.
Explore Workflows Documentation →
Launch and Track Outbound Campaigns
You can now build agents that initiate calls. We’ve shipped a new set of /campaign endpoints to let you launch, track, and manage outbound campaigns directly via the API. Monitor real-time call states like queued, in-progress, or voicemail-detected to measure performance and build automated follow-up logic.
This unlocks use cases like proactive appointment reminders, automated lead outreach, and batch customer notifications.
Read the Campaigns Docs →
Bring Vapi Agents to the Browser with Web Chat
The Vapi platform now supports text. We’ve launched a new OpenAI-compatible web chat endpoint that allows you to deploy your assistants in any browser. It supports streaming, can be stateless or session-based, and lets you reuse the same agent logic you’ve already built for voice.
It’s perfect for building hybrid apps, support widgets, lead capture forms, or internal tools.
Explore Web Chat Docs →
🤖 New Models, Transcribers & Voices
We’re constantly expanding the stack of tools you can build with. This month’s additions include:
- Inworld: Now integrated as a voice provider. Supports EN, CN, KR with voice caching and auto-detection.
- Cartesia Ink Whisper: A new real-time transcriber option delivering high-accuracy streaming at under 100ms.
- Gladia Solaria-1: Added support for this model, offering multi-language audio transcription with enhanced quality.
- Multilingual LMNT Voice: You can now set languages using ISO 639-1 codes or enable auto-detection for LMNT voices.
Check your Vapi Dashboard to explore these options.
🛠️ Developer Tools & API Enhancements
- Granular Cost Observability: Each call and chat object now returns a cost breakdown for LLMs, voices, transcription, and transport, giving you full visibility. [Docs]
- Filter Call Logs by Phone Number: You can now filter call logs directly by the customer's phone number in the dashboard for faster debugging. [Dashboard]
- Warm Transfer Upgrades: Warm transfers now support configurable timeouts and optional hold music for a better user experience. [Docs]
- SIP Call Transparency: Shipped more specific endedReason codes for SIP failures and caller disconnects, giving you more precise control over retry logic. [Docs]
- HMAC Webhook Auth: We’ve added a simpler, more secure method for webhook verification. [Docs]
- Azure OpenAI Mode: Vapi now automatically strips unsupported parameters when you use Azure-hosted models, preventing common errors. [Docs]
🎉 Community & Events
-
VapiCon 2025: We’re hosting our first developer conference this October in San Francisco. It’s a full day focused on real products, real voice use cases, and real builds.
Early bird tickets now available →
-
Vapi Build Challenge (Concluded): Our 45-day build challenge wrapped up with over 300 incredible submissions from the community. A huge thank you to everyone who participated. See the winning projects →
-
SF Voice AI Builders Meetup (New): Join us for the launch of the Bay Area's first dedicated Voice AI Builders Meetup in San Francisco on July 18. Show off what you're building, learn from others, and connect with experts.
Request to Join →
⚡ Breaking Changes & Deprecations
To keep our API clean and performant, we've made a few updates.
- Removed: codeSwitchingEnabled (Deepgram), aws-sts (OAuth), and regex support in several schema fields.
- Deprecated: backgroundDenoisingEnabled is now replaced by the more flexible backgroundSpeechDenoisingPlan.
- Sunsetting: PlayHT voice support will be removed from the dashboard on July 26, 2025.
View the full changelog →
What are the next big hurdles you're facing in production or development?
Where should we focus our efforts next?
Let us know in the Discord or here.
As always, we're eager to see what you build
