About the Project
What Inspired Us
Social media is one of the most powerful communication tools today—but managing it is time-consuming, repetitive, and fragmented. We wanted to create something simple, intuitive, and truly helpful: a voice-first AI agent that could handle social media the same way a personal assistant would. Inspired by how we naturally speak about our content ("Post that picture I took last night"), we imagined an app that could understand that intent and make it happen—across multiple platforms—with zero friction.
What We Built
We built EchoPost, a mobile app that lets users post to Facebook, Instagram, YouTube, Twitter and TikTok using only their voice. From selecting the right photo based on metadata (like time or location) to formatting captions and publishing posts via platform APIs, Echo Post streamlines the entire content publishing process into a single voice command.
How We Built It
- Frontend: Flutter (for cross-platform mobile support on iOS and Android)
- Voice Recognition: OpenAI Whisper
- Prompt-to-Action Mapping: Custom GPT logic via Bolt AI, which transforms voice commands into structured JSON task plans
- Media Matching: Metadata filters to locate images and videos based on natural language descriptions
- API Integration: OAuth for supported social platforms (starting with Twitter and Facebook for demo reliability)
- Cloud Backend: Firebase for user auth, media indexing, and token management
What We Learned
- How to build an agentic AI system that translates human intent into precise, platform-specific actions
- How to blend voice interaction, NLP, and social media APIs into one cohesive UX
- How much people actually want this: during testing, nearly everyone who saw a demo asked, “Can I use this now?”
Challenges We Faced
- Media Matching: Selecting the right photo based on vague commands like “that one from last Friday” pushed us to rely on metadata and voice clarifications
- API Restrictions: Instagram and TikTok have limited or business-only APIs, so we built workarounds and simulated posting flows for the demo
- Voice Ambiguity: Human speech is wonderfully flexible—too flexible for a naive parser. Bolt AI gave us structured predictability from natural input.
Meet the Team
Ozie Cargile is a visionary in artificial intelligence and music production, blending deep learning with creative expression under the banner of Espressivo. His background spans orchestral composition, Forex market prediction, and AI-integrated software. Contact: ocargile@gmail.com
Louis Brent, known as the API King, brings unmatched expertise in web and mobile app development. With a track record of seamless API integrations across platforms like Facebook, Twitter, Instagram, and OpenAI, Louis makes complex systems feel intuitive. Contact: louisbrent1992@gmail.com
The two met in 2020 and have been collaborating ever since—combining Ozie’s AI and creative vision with Louis’s technical execution. Together, they’ve built EchoPost, a voice-powered social media agent app that showcases the best of both worlds.
Built With
- android-studio
- boltai
- chatgpt
- claude
- cloud-storage
- dart
- facebookgraphapi
- firebase
- firebaseauthentication
- firebasecloudfunctions
- firestore
- flutter
- github
- javascript
- oauth
- openaiwhisper
- tiktokapi
- whisper
- xcode
Log in or sign up for Devpost to join the conversation.