This July, we focused on making Vapi more accessible to developers at every level. From a brand new CLI to embeddable widgets, we shipped tools to help you go from idea to production faster, regardless of your stack.
Here's everything we shipped this month:
New Tool
Build and Deploy from the Command Line with the Vapi CLI
We've launched the Vapi CLI, your new command-line companion for building, configuring, and managing voice assistants. Now you can scaffold a new Vapi project with a single command, saving you setup time so you can focus on building your agent's logic.
The CLI has native support for React, Vue, Next.js, Python, Go, and Flutter, and integrates with developer environments like Cursor, Windsurf, and VSCode.
Explore the CLI →
Embed Voice AI Into Your Product with Voice & Chat Widgets
You can now create and embed customizable Voice or Chat Widgets directly from your Vapi Dashboard. This is the fastest way to add a voice interface to your existing product or website.
What you can now do:
- Full Customization: Control the widget's appearance, size, text, and layout to match your brand.
- Mode Selection: Let users choose between voice and chat interactions.
- Compliance Built-In: Easily configure and display consent banners for data privacy.
Read the Widget Docs →
New from Vapi
Introducing Voice AI Weekly: The Newsletter for Builders
The voice AI space moves fast. To help you cut through the noise, we've launched Voice AI Weekly.
Every week, we deliver the top 3-5 stories that actually matter for your roadmap, spotlight a real company building with Vapi, and provide a scannable roundup of AI news. No fluff, just signal.
Subscribe to the newsletter →
Platform Updates
Deeper Control Inside Workflows
We’ve added more granular tools to help you build more sophisticated conversational logic.
- Conversation Node Tools: Add tools directly to conversation nodes or reference existing tools by ID for more modular and reusable logic.
- Custom Background Sounds: Configure custom background audio beyond the default office sounds to create more immersive call experiences.
- API Response Extraction: Use our new UI builder to easily extract and map variables from API responses directly within a workflow.
- Keypad Input Support: Configure keypadInputPlan to build interactive menus and capture DTMF input during calls.
Explore Workflows →
🤖 New Models, Providers & Integrations
- Minimax Voice Provider: Added a new voice provider with customizable pitch, speed, and emotion controls.
- AssemblyAI Transcriber Updates: We've switched to the new Universal Streaming API by default for faster, more efficient transcription. Configuration is now simpler with formatTurns defaulting to true. [Read the docs]
Check your Vapi Dashboard to explore these options.
🛠️ Developer Tools & API Enhancements
- Full CRUD in API Request Tool: The tool now supports PUT, PATCH, and DELETE methods, enabling full CRUD operations. [Docs]
- Custom LLM Headers: You can now add custom HTTP headers to Custom LLM Models and Workflow Custom Models. [Docs]
- Workflow Call-Level Hooks: Define call-level logic with new hook actions like say, tool, transfer, and function. [Docs]
- Enhanced Chat Events: We’ve shipped new message types for chat.created, session.updated, and more for better state management. [API Reference]
🔧 Stability & Performance Updates
We're continuing to harden our infrastructure for production workloads.
- Dashboard Stability: Shipped fixes to reduce "fail whale" occurrences and improve stability across dashboard pages.
- Billing Bug Fixes: Addressed an issue where charging errors could occur during plan downgrades.
- LLM Stability: Moved GPT-4o back to the completions API to reduce repetition and hanging issues.
- Enterprise Tracking: Improved payment and invoice tracking for Enterprise customers.
🎉 Community & Events
- VapiCon 2025: We're hosting the Bay Area's first-ever Voice AI summit this October in San Francisco. It’s a full day focused on real products, real use cases, and real builds.
Early bird tickets are ending soon →
- SF Voice AI Builders Meetup: Join us for our August meetup on Friday, August 22. It's a monthly gathering for the Bay Area's top voice AI builders to share, learn, and connect.
Request to Join →
⚡ Breaking Changes & Deprecations
- Removed Models: gpt-4.5-preview, o1-preview, and o1-preview-2024-09-12 are no longer available.
- AssemblyAI: The enableUniversalStreamingApi property is now deprecated from the transcriber configuration as it is enabled by default.
View the full changelog →
What are you building with these new tools? What should we focus on next to help you ship faster?
Let us know in the Discord or on our feature requests board.
Happy building.
