Inspiration

Working with LLMs often feels like guesswork. You spend time tweaking prompts, switching models, adjusting settings — yet you're never sure if you're getting the best possible output. We wanted to simplify this process and bring structure to LLM interaction, just like Zeno of Elea brought logic to paradoxes. Zeno helps eliminate prompt chaos and model confusion — so builders can focus on results, not trial-and-error.

What it does

Zeno is an intelligent model refinement and inference optimization tool. It helps users:

  • Rewrite and optimize their prompts for better LLM performance
  • Get smart model recommendations based on their task
  • Control output quality with simple and advanced settings
  • Optionally, compare outputs across models to find the best result

It's designed for technical teams, AI builders, and anyone working with LLMs who want less guesswork and better results.

How we built it

We used:

  • Bolt.new for the landing page and initial interface design
  • Custom frontend for the logged-in dashboard with prompt refinement and settings
  • Integrated multiple LLM APIs (e.g., GPT-4o, Claude 3.5) for real-time prompt testing
  • Built a logic layer for prompt classification, model suggestion, and advanced control
  • Designed with a clean, modern, terminal-inspired interface for a unique user experience

Challenges we ran into

  • Balancing simplicity for beginners with deep control for advanced users
  • Creating a design that feels different from typical SaaS layouts but remains usable
  • Ensuring model recommendations are helpful without being restrictive
  • Keeping the product focused within the limited time of the hackathon

Accomplishments that we're proud of

  • Developed a working prototype of the full prompt-to-output workflow
  • Created a unique, clean interface that feels distinct from existing AI tools
  • Built a flexible system supporting both quick-start and advanced use cases
  • Solidified Zeno's core identity as more than just a prompt tool — it's an AI interaction optimizer

What we learned

  • Good prompt engineering isn't enough — model choice and output tuning matter just as much
  • Simple, thoughtful design can dramatically improve user experience, especially in technical tools
  • Hackathon time pressure forces focus — scoping the right features is critical
  • Building AI tools for technical users requires both power and restraint in design

What's next for Zeno

  • Expanding model support beyond initial LLMs
  • Building deeper output comparison tools
  • Adding prompt history, versioning, and collaboration features
  • Exploring cost optimization suggestions for LLM inference
  • Launching publicly for AI teams and builders who want to refine their workflows

Built With

Share this project:

Updates