Forest background

Uncapped - AI chat app. Choose between popular AI models and use your own API key

Date: 1/2/2025

Services: Complete website, from design to development

Main image for Uncapped - AI chat app. Choose between popular AI models and use your own API key

About Uncapped

Uncapped is a privacy-focused AI chat app. It allows users to choose between popular AI models and use their own API key to save money.

Uncapped runs fully locally, meaning there is no server-side processing. It uses IndexedDB (browser storage) to store chat history and custom chat profiles.

Built-in memory

One of the main and differentiating features of Uncapped is the built-in memory.

For every message you send, it is sent to a cheaper memory model, like GPT-4o-mini, along with a prompt that determines if it should be stored in memory or not.

If it should be stored, the message is added to the memory and sent with every future prompt. This way, the AI will remember details about you and your conversations.

Uncapped Memory

Local database using IndexedDB

Every message you send is stored in browser storage (IndexedDB). Uncapped works locally, meaning every message is directly sent to the AI model.

With other AI apps, a message is first sent to the server, which in turn sends it to the AI model. This introduces unnecessary latency and has an impact on privacy.

Custom chat profiles

Uncapped also allows you to create custom chat profiles and attach default context to them.

For example, you can create a chat profile for your business and attach some default context to it (business location, software you use, etc). Now when sending a message, the AI will know about your business and can use that information to more accurately answer your questions.

Uncapped Chat Profiles

Choose between 30+ built in themes

Uncapped was built from the ground up to allow users to choose between 30+ built in themes. Each with different color schemes, borders, and overall look and feel.

Uncapped Themes Uncapped Themes Uncapped Themes

Adapts to multiple LLM providers

From a technical standpoint, Uncapped is built to be very flexible and adapt to multiple LLM providers.

When utilizing an LLM provider as developer, you usually have to use their API endpoint that streams chunks of data. Each chunk usually contains a message and a token count.

However, these chunks can be very different from one provider to another. Uncapped was built with a provider agnostic approach using the adapter design pattern.

This way, the code is very flexible and can be easily extended to support new providers.

Technologies Used:

  • Next.js
  • Typescript
  • Tailwind CSS (with DaisyUI)
  • PWA
  • IndexedDB
  • Deployed on custom Hetzner server