Inspiration
We wanted to save the world. And started with use cases that overwhelmed us. We had a hard time narrowing down our ideas, and figuring out how prompt-chaining would solve that issue.
Eventually we honed in on one intent that we felt good about: interview prep. We wanted to build a bot that could help users practice interview questions for future roles, and suggest improvements to their answers. 5.8 million are currently unemployed in the US, and we feel that this project would be helpful for job seekers.
What it does
In Snoop Dogg’s voice and tone, it helps people prepare for interviews and improve their interview questions. When a user answers an interview question, it evaluates what was successful about the response, gives recommendations on how to improve the response, and rewrites the original answer applying those suggestions.
We used Snoop Dogg for demo purposes, but you can change the person to anyone you like.
How we built it
We used Voiceflow to build the bot, plus ChatGPT and GPT-4.
Challenges we ran into
Ambitious scope given the time we had to build the project Nailing the voice and tone (and output) through prompt engineering. We’re conversation designers and writers, so we judged the output’s quality with that lens.
Accomplishments that we're proud of
We’re proud that we created a chatbot that’s creative and useful–and accomplishes that with clever prompt engineering of ChatGPT and GPT-4.
What we learned
We learned that the slightest change in prompt phrase altered the entire flow. After a hundred replays, and "why is the persona happening here and not there?!", we arrived at a coherent conversation flow that was equal parts helpful and entertaining. After 6 hours of trying to reword prompts and re-watching Denys' videos, we managed to tease out the experience we were looking for.
What's next for Interview Prep Bot
We initially envisioned this as a full-fledged job search experience that would handle generating a new resume and cover letter, AI headshot, career advice, and helping a user upskill. In the short-term, we’d like to add an ASR component so that the user can speak their response and interact with voice. This would also allow sentiment analysis and the ability to provide more meaningful feedback.
Built With
- chatgpt
- gpt-4
- voiceflow



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