Video Voicemail

Personalized Video Outreach at Scale. Send thousands of personalized video messages via SMS in minutes.

Comment

Inspiration

The inspiration for Video Voicemail came from the challenge of standing out in today’s crowded outreach landscape. Traditional email and cold calls just aren’t getting results—they’re ignored, filtered, or simply forgotten. But personalized video? That gets attention. I wanted to make it easy for businesses to scale personal, face-to-face outreach—using AI to create highly customized video messages for every lead or customer, without hours of manual effort.

We saw a chance to blend the power of synthetic video, AI voice, and automation to let anyone send tailored video messages at scale—boosting open rates, engagement, and conversion in a way that text simply can’t compete with.

What it does

Video Voicemail is a SaaS platform that lets you generate, personalize, and deliver short AI-powered video messages to any lead list—at scale. Here’s how it works:

  • Personalized Video Generation: Enter a business description or campaign goal, and the system programmatically creates 3–5 scene prompts.
  • Visual Content Creation: Each prompt is turned into short video scenes using Luma, then stitched together to create a dynamic full video.
  • Dynamic Intro: a 2-second personalized intro is created with Tavus, greeting each recipient by name or referencing their business.
  • Video Stitching: All the scenes (and intro, if included) are stitched together into a seamless video using Shotstack.
  • Voice Over: Add a human-like AI voiceover (male or female by default, with custom voices supported soon) using ElevenLabs.
  • Thumbnail Generation: Each video gets a branded, eye-catching thumbnail.
  • Delivery: The final video is sent via SMS using Twilio, or can be embedded in emails or landing pages.

No video editing required, no technical skills needed—just scalable, truly personal video outreach in a few clicks.

How we built it

  • Frontend: Built on Bolt.new with a focus on simple, guided flows and client branding.

  • Backend Integrations:

    • Supabase for authentication, storage, and state management.
    • Luma API to generate high quality video.
    • Tavus API for fast, hyper-personalized avatar intros.
    • Shotstack API for robust, programmatic video stitching and rendering.
    • ElevenLabs API for natural, customizable voiceovers.
    • Twilio API for automated video SMS delivery.
    • Cloudinary/Pica for thumbnail generation image processing and video upliads.

  • Logic: When a user submits a business description, we break it into scenes, generate assets, stitch and render the video, create a thumbnail, and deliver—all orchestrated via backend Bolt functions. State is tracked in Supabase for real-time status updates.

Challenges we ran into

  • API Headaches: Chaining together multiple third-party APIs (Runway, Shotstack, Tavus, ElevenLabs, Twilio) led to some gnarly error-handling, race conditions, and asynchronous headaches.
  • Authorization & CORS: Bolt.new’s edge functions required careful handling of secrets and CORS, especially with Supabase and third-party API calls.
  • Video Stitching: Stitching short video clips into a seamless, branded experience (with voiceover in sync) took a lot of trial and error—especially with variable render times and Shotstack’s polling system.
  • Fallback Logic: Handling failures gracefully—like if a video failed to render or a thumbnail couldn’t be generated—so users always got a result.
  • Cost Management: Managing API usage (some providers bill per second or per request) while keeping the platform affordable for users.

Accomplishments that we're proud of

  • Automated, End-to-End Video Creation: Users can create multi-scene, personalized videos at scale in under a minute.
  • Seamless Third-Party Integration: Successfully connected 5+ APIs with smooth user experience and robust error handling.
  • Personalization at Scale: The Tavus intro and dynamic script generation mean every video feels tailor-made.
  • Usability: The frontend is clean and simple—no video editing skills required. Anyone can use it.
  • Stability: Built-in fallback logic ensures a finished video, even if one step fails.

What we learned

  • APIs are powerful, but brittle: Every integration adds risk, and robust error handling is critical.
  • Speed matters: Even a 10-second delay can feel long—optimizing render and polling times is huge for user satisfaction.
  • Personalization wins: Videos that greet people by name or reference their company crush generic outreach, every time.
  • Documentation is your friend: Keeping good docs for every API, workflow, and fallback saved hours of debugging.

What's next for Video Voicemail

  • Custom Avatars & Voices: Supporting fully branded avatars and user-trained voices for true white-label outreach.
  • Team Features: Multi-user workspaces, campaign analytics, and deeper CRM integrations.
  • Omni-channel Delivery: Expanding beyond SMS to email, LinkedIn, WhatsApp, and more.
  • A/B Testing: Built-in split testing to see which video styles or scripts convert best.
  • Marketplace for Creatives: Letting agencies and freelancers offer video campaigns to clients through our platform.
  • Cost Optimization: Smarter asset caching, queueing, and provider selection to keep pricing competitive as we scale.

Built With

Share this project:

Updates