Verba

Inspiration: I use AI to help me write, but I often find that the output sounds polished yet generic — it doesn’t sound like me. The problem is not that AI cannot write. The problem is that AI does not understand how a specific person thinks, structures ideas, and expresses opinions. I wanted to build a tool that helps AI preserve a person’s unique expression style instead of producing another generic AI-generated response.

What it does: Verba analyzes a user’s previous writing samples and builds a personal expression profile based on their writing patterns, including:

how they start a topic; how they structure arguments; how they use examples; their sentence rhythm; their tone and communication style. Users can then use Verba to transform generic AI-generated content into content that better matches their own expression style.

How we built it: I first designed the product framework and style analysis system with ChatGPT, defining how personal expression could be broken down into measurable dimensions. Then I used Codex to implement the prototype, build the application, and iterate quickly from idea to a working product. The development process combined product thinking, linguistic analysis, and AI-assisted coding.

Challenges we ran into: The biggest challenge was that early outputs were technically correct but still felt generic. The system could rewrite text, but it did not truly capture personal expression.

To solve this, I repeatedly reviewed the results, refined the style framework, adjusted the analysis dimensions, and improved the logic with Codex until the output became more aligned with the intended personal style.

Accomplishments that we're proud of: We are proud that we built a working prototype that goes beyond simple AI rewriting. Instead of only changing wording, Verba attempts to model deeper expression patterns, such as: personal communication habits; thought organization; writing rhythm; preferred ways of explaining ideas. The project demonstrates how AI can become more personalized by understanding the user, not just generating better text.

What we learned: We learned that making AI output feel personal is much harder than making AI output sound fluent.

A person’s writing style is not defined by a few favorite words. It comes from deeper patterns: how they frame problems, build arguments, choose examples, and communicate uncertainty. Building a personalized AI experience requires both technical implementation and a strong understanding of human communication.

What's next for Verba: The next step for Verba is to evolve from a writing transformation tool into a personal expression model. Future versions could help users maintain their own voice across different contexts, including: social media posts; emails; articles; presentations; professional communication.

Our long-term vision is to build an AI assistant that does not just write for you, but understands how you express yourself.

Built With

• chatgpt
• codex