Foligo

• 

• 

• 

• 

• 

• 

• 

• 

• 

  App Icon

Foligo: Your Conversational Portfolio Assistant

Foligo is an AI-powered platform that generates polished portfolio content, blog posts, and social media updates from a simple verbal description of your projects. Go from an idea to a live portfolio piece in minutes, just by talking.

The Problem

Students and professionals often lack the time to sit down and write detailed descriptions of their projects and accomplishments for their portfolios. This friction results in outdated websites and missed opportunities to showcase their work. There’s currently no solution that allows for productive, on-the-go content creation, meaning valuable experiences are often forgotten or poorly documented.

The Solution

Foligo is an application that allows users to create professional content simply by talking. We were inspired by the idea of turning dead time, like walking between classes, into productive portfolio-building sessions. A user starts a conversation with our AI assistant, which asks intelligent, clarifying questions to gather details. From that conversation, Foligo instantly generates a detailed portfolio write-up, a professional LinkedIn post, and a concise update for platforms like X (formerly Twitter).

Our Technology

We used Next.js and React with Tailwind CSS to develop the responsive web application. The backend is built on Node.js and Express.js, and it interfaces with our PostgreSQL database via a custom RESTful API. We integrated the Google Gemini API to power our conversational engine and content generation, and used the ElevenLabs API for high-quality speech-to-text and text-to-speech, enabling the core voice-driven experience.

Challenges and Solutions

Developing a multi-turn conversational AI that maintains context and asks relevant, non-repetitive questions was a significant challenge. We had to carefully engineer the prompts for the Gemini API through extensive testing and refinement. Going into this challenge, we planned to create our own speech to speech agent using Vultr, Digital Ocean, and open source speech models and LLMs. However, we were not allowed to use any GPUs from either AI company and spent hours getting support only to be told we cannot use any high compute resources. Despite this setback, we still wanted to use Vultr and Digital Ocean so we hosted portions of our site on their CPU servers using a LAMP stack and pivoted to the ElevenLabs API for real-time, low-latency voice interaction. We worked to manage the audio streams effectively to ensure a smooth, natural-feeling conversation without awkward delays.

Accessibility

Our voice-first design is central to our commitment to accessibility. By allowing users to create content entirely through conversation, Foligo provides a powerful alternative to traditional keyboard-and-mouse interfaces. This makes portfolio creation accessible to users with motor impairments or those who find extensive typing difficult. The clean, structured UI of the generated websites also adheres to best practices for screen reader compatibility.

Accomplishments that we’re proud of

We are proud of creating a fully functional, end-to-end application that solves a real-world problem for students and professionals. Successfully integrating multiple complex APIs (Gemini for intelligence, ElevenLabs for voice) to create a single, seamless user experience feels like a major accomplishment. We are also proud that our platform can generate three distinct, high-quality pieces of content for different platforms from a single conversation.

What we learned

We learned the nuances of prompt engineering for large language models like Gemini to control the tone, structure, and detail of generated text. Our team also gained valuable experience with the practical challenges of implementing real-time voice AI and building a robust RESTful API to support a complex, stateful workflow.

What's Next for Foligo

We plan to add more features to the platform, such as AI-powered image selection that can analyze a project URL, take screenshots, and select the most relevant images for a post. We also envision an interview preparation tool that leverages a user's accumulated project data to act as a personalized prep assistant for specific job interviews.

Built With

• digitalocean
• elevenlabs-api
• express.js
• gemini-api
• jwt
• next.js
• node.js
• react
• rest-api
• vultr