Goggins AI

⭐ Goggins AI — Stay Hard. Stay Focused. Stay Winning.

Inspiration

We all procrastinate. Hours of potential get lost to YouTube spirals, Discord chats, and mindless scrolling. We wanted a way to bring the intensity of David Goggins directly into our workflow, someone who literally watches your screen and calls you out whenever you become "soft".

The next generation of motivation isn't a quote on your wall. It's an AI coach staring you down while you work.

What it does

Goggins AI watches your screen (literally, Goggins' face is on your screen), tracks your productivity, and speaks to you, helping you realign with your goals and calling you out when you're distracted.

How we built it

We combined multiple AI pipelines to create a seamless motivational experience: Periodic Screenshots → contextualized by GPT-4o-mini for real-time understanding Productivity Analytics → generated using GPT-4o Voice Interaction →

Whisper for transcription
ElevenLabs for text-to-speech (yes, it sounds exactly like Goggins) Tech Stack → JavaScript & TypeScript, custom desktop overlays, and local model experimentation We experimented with locally hosted models (LLaVA, Ollama 2.3, Qwen3, Whisper) to compare latency vs. quality and choose the optimal setup for speed + accuracy.

Challenges we ran into

The hardest part was balancing conversational flow with productivity insights while keeping Goggins responsive. To resolve this, we tested multiple locally hosted models (LLaVA, Ollama 2.3, Qwen3, Whisper) to find the best balance of speed and quality. Goggins' voice would get recaptured as the user's own voice input, so the mic had to be disabled as Goggins is talking.

Accomplishments that we're proud of

It sounds exactly like Goggins. It wakes you up and helps you stay focused, and genuinely feels like interacting with a coach who wants you to win.

What we learned

It's difficult for the transcribing model to know when you've stopped talking, and merging all our commits into the same branch makes it difficult to manage, locally running isn't always faster (depends on your compute)

What's next for Goggins AI

We're going to make it into a product and let the world experience next-level productivity. AI can make us 10x engineers. Goggins AI makes us 10x humans.

https://memorable-artboard-111910.framer.app/goggins-ai-landing