AI Career Guide: Voice-Powered Bedford Assistant

An AI voice assistant that guides students through Bedford's career counseling process using natural conversation

Comment

Inspiration

Inspiration I've always been fascinated by the idea of making career guidance more human and accessible—especially for students and professionals who may not know where to start. Bedford's career model is powerful but often feels static and impersonal. The emergence of real-time conversational AI like ElevenLabs and GPT-4o inspired me to give Bedford a voice—literally.

What it does AI Career Guide is a voice-enabled assistant that helps users understand and navigate the Bedford Career Counselling model. Users can ask questions out loud, receive intelligent, context-aware guidance, and interact naturally—like having a conversation with a real career coach.

How I built it I combined ElevenLabs' Conversational AI (for voice interaction and TTS) with OpenAI's GPT-4o (for reasoning and natural language understanding). I designed a custom system prompt to give the assistant a friendly yet professional tone. Using Python and Flask, I built a backend that connects the user’s voice input to the LLM, processes responses, and returns voice output. The frontend was built with HTML, CSS, and JavaScript for a lightweight and intuitive experience.

Challenges I ran into One of the toughest challenges was tuning the system prompt to balance professionalism with friendliness while still remaining aligned with the Bedford framework. Another challenge was managing real-time latency between voice inputs and model responses to keep the interaction fluid. Choosing the right temperature and model variants for a conversational experience also took experimentation.

Accomplishments that I'm proud of I’m proud of building a seamless voice interface that makes structured career guidance feel personalized and engaging. Watching the assistant respond naturally and help users explore their career paths through conversation was a rewarding moment.

What I learned I learned how to fine-tune LLM behavior using system prompts, how to integrate ElevenLabs for real-time voice interactivity, and how to orchestrate a smooth experience between multiple APIs. I also gained a deeper appreciation for conversation design and user experience in voice-based interfaces.

What's next for AI Career Guide: Voice-Powered Bedford Assistant Next, I plan to:

Add multilingual support for broader accessibility

Integrate voice memory so the assistant remembers user context across sessions

Expand it to include resume feedback and mock interviews

Package the assistant as a widget for educational platforms and career portals

What it does

How I built it

Challenges I ran into

Accomplishments that I'm proud of

What I learned

What's next for AI Career Guide: Voice-Powered Bedford Assistant

Built With

  • ai
  • and
  • api
  • applies)
  • backend
  • build
  • built
  • conversational
  • data
  • deploying
  • dialogue
  • dynamic
  • elevenlabs
  • exchange
  • flask
  • for
  • framework
  • frontend
  • frontend/backend
  • generation
  • gpt-4o
  • handling
  • html/css
  • integration
  • intelligent
  • interaction
  • interactions
  • interface
  • javascript
  • json
  • language
  • lightweight
  • natural
  • openai
  • or
  • postman
  • prompt
  • python
  • real-time
  • render
  • requests
  • scripting
  • speech
  • structuring
  • style
  • synthesis
  • testing
  • the
  • to
  • understanding
  • use
  • user
  • validating
  • vercel
  • voice
  • whichever
  • with:
Share this project:

Updates