Self-hosted LLM chat with local/API model mixing

Problem Context

Developers and technical teams rely on LLM chat apps for coding, research, and collaboration. They need a self-hosted solution to avoid corporate bloat, mix local models with API-based models, and extend the code easily. Current tools either lack these features or are too complex.

Pain Points

Users waste time switching between local and API-based models manually. Existing apps are bloated with unnecessary features or unfinished functionality. They struggle to find a simple, extensible tool written in TypeScript/Go that meets their needs without corporate overhead.

Impact

Poor chat apps slow down workflows, leading to lost productivity and frustration. Teams spend hours configuring or working around limitations instead of focusing on their core tasks. The lack of a lightweight, flexible solution forces users to accept subpar tools or build their own from scratch.

Urgency

This problem is urgent because teams cannot afford downtime or inefficiencies in their daily workflows. A poor chat app directly impacts coding speed, research accuracy, and collaboration quality. Users need a reliable, self-hosted solution now to avoid continued productivity losses.

Target Audience

Developers, data scientists, and technical teams who self-host tools and need a flexible LLM chat for coding, research, or collaboration. This includes open-source contributors, AI researchers, and small tech companies that prioritize control and customization over convenience.

Solution Approach

A self-hosted LLM chat app that lets users mix local models (e.g., Llama, Mistral) with API-based models (e.g., OpenAI) in one interface. Built in TypeScript/Go for easy extension, it avoids corporate bloat and focuses on simplicity and developer workflows. Users pay for API model support or premium features.

Key Features

1. Lightweight Codebase: Written in TypeScript/Go for easy customization and minimal dependencies.
2. No Bloat: Stripped-down interface with only essential features (e.g., no telemetry, no forced updates).
3. Extensible Plugins: Add-ons for coding (e.g., GitHub integration) or research (e.g., data analysis tools).

User Experience

Users install the app locally, configure their preferred models, and start chatting. They switch between local/API models with one click, carry over context, and extend the chat logic via TypeScript/Go. The app integrates into their existing workflows without disrupting productivity.

Differentiation

Unlike bloated corporate apps or incomplete open-source tools, this solution focuses on simplicity + extensibility. It’s the only self-hosted LLM chat that natively supports mixing local/API models while being lightweight and developer-friendly. Competitors either lack model flexibility or are too heavy for self-hosting.

Scalability

Starts as a single-user self-hosted tool, then adds team features (shared chats, model pools). Expands with plugins (e.g., coding/research tools) and a marketplace for custom model adapters. Revenue grows via per-seat pricing and premium model support.

Expected Impact

Users regain productivity by eliminating manual model switching and bloated features. Teams collaborate more efficiently with a chat app tailored to their workflows. The app becomes a mission-critical tool for developers and researchers who need reliability and customization.