Inspiration

OmniVeo: The AI That Learns How You Learn

The Moment Everything Changed

Picture this: It's 2 AM, and I'm staring at Photoshop, completely defeated. I'm a junk hauler who fell down the AI rabbit hole three years ago, and now I'm trying to design skateboard decks for our small business. Hours of YouTube tutorials later, I'm still lost.

That's when it hit me: What if AI could see what I'm doing and teach me in real-time?

Five weeks later, that midnight frustration became OmniVeo—the world's first AI tutor that watches your screen and guides you as you work.

The Breakthrough

OmniVeo isn't just another tutorial platform. It's your personal creative mentor that:

  • Sees your screen in real-time (screenshots every 3 seconds)
  • Speaks to you naturally through advanced voice AI
  • Adapts to your learning style and skill level
  • Never breaks your workflow - no switching between apps

From Photoshop to Maya, Premiere Pro to Figma—across 12 powerful tools in 6 creative categories—OmniVeo transforms how people master creative software.

The Team That Made It Happen

The Pivot That Changed Everything

Day 18 of the Bolt.new Hackathon: Our original concept "Forge" was too ambitious. I put out a desperate call on Discord for help, and magic happened.

Maruf answered the call first. With full-stack mobile experience under his belt, he didn't just join—he revolutionized our backend. He breathed life into ElevenLabs integration, connected Gemini Vision, and built custom RAG pipelines from Pinecone. But here's the kicker: he took our "vibe coding" to the next level by adapting features from demo repos that we never thought possible.

Tobi brought the soul to our system. His UI/UX expertise and branding vision created designs that don't just function—they speak to users. Every pixel tells a story, every interaction feels intentional. When users see OmniVeo, they immediately understand its power.

Swara became our innovation catalyst. Her technical prowess turned wild ideas into working reality. She didn't just code—she architected solutions, provided crucial feedback, and constantly pushed us to think bigger. Her smaller experimental features became the foundation for our most impressive capabilities.

The Technical Magic

Here's how we built an AI that truly gets you:

The Stack

  • Bolt.new - Rapid frontend development
  • Supabase - Authentication & user management
  • ElevenLabs - Conversational AI that feels human
  • Google Gemini Vision - Real-time screen analysis
  • Pinecone - Smart knowledge retrieval
  • n8n - Seamless workflow automation

The Process (The Real Innovation)

  1. User grants screen + mic access → Learning session begins
  2. Screenshots captured every 3 seconds → Sent to n8n workflow
  3. Gemini Vision analyzes context → Returns structured JSON summary
  4. Pinecone searches relevant knowledge → RAG-powered context matching
  5. ElevenLabs delivers personalized guidance → Real-time, contextual help

The result? An AI that doesn't just see what you're doing—it understands what you're trying to achieve.

The Battles We Fought (And Won)

Documentation Nightmare

Maya and DaVinci Resolve documentation was scattered and inconsistent. Solution? We built our own knowledge base through community-driven content and real-world usage patterns.

The Token Wars

ElevenLabs and Gemini APIs had strict limits. We optimized every prompt, compressed data streams, and maintained natural conversation flow despite technical constraints.

The 8-Day Sprint

After pivoting from Forge, we had just 8 focused days. Non-negotiable deadline, four-person team, multiple time zones. We made it work through radical focus and clear role division.

Leading Without Coding

As a non-technical founder, I learned that vision and strategy matter as much as implementation. My job became connecting user needs to technical possibilities.

What We Achieved (In Just 3 Weeks)

Built the impossible - Real-time AI tutoring with screen awareness and voice interaction

Formed a dream team mid-hackathon and found perfect synergy

Created beautiful UX that makes complex AI feel simple and intuitive

Stayed true to our mission - Making powerful creative tools accessible to everyone

The Bigger Picture

We discovered something profound: Voice AI isn't just the future—it's the present. When an AI remembers who you are, adapts to how you learn, and guides you through real work, it stops being a tool and becomes a companion.

OmniVeo proves that AI should empower individuals, not replace them.

What's Next: The Revolution Continues

Immediate Roadmap

  • Expand tool support across more creative categories
  • Enhanced memory systems for deeper personalization
  • Beta program launch with real creators and educators
  • Partnership network for continuous improvement
  • Sustainable pricing model that prioritizes access

The Vision

OmniVeo will become the standard for learning creative software—not through passive watching, but through active doing, with an AI mentor that truly knows and supports you.

Because the best way to learn isn't to watch someone else create—it's to create alongside someone who believes in your potential.


Team Values: We believe AI should empower individuals and make them better, not replace them. Every feature we build serves this mission.

What it does

How we built it

Challenges we ran into

Accomplishments that we're proud of

What we learned

What's next for OmniVeo

Built With

  • elevenlabs
  • pinecone
  • rag
  • react
  • supabase
  • vite
Share this project:

Updates

posted an update

OmniVeo is evolving!

We’re currently reengineering the backend to make OmniVeo more affordable and unlock unlimited access for everyone. No more tab-switching or falling into YouTube rabbit holes — this is what the future of learning complex software should feel like.

For the Bolt hackathon, we didn’t even get the app fully functioning until the morning of submission. We had just 1 hour to put together the demo video… and ended up submitting with only 8 minutes left on the clock. It was a real sprint to get the MVP across the finish line — but we made it.

The production version will be far more polished, stable, and powerful than what we submitted. Expect a beta within the next 30 days, and stay tuned — a lot more is coming soon.

Log in or sign up for Devpost to join the conversation.