Twigg

Twigg is a context management platform designed for long-term AI projects. It rethinks the traditional chat interface by providing an interactive tree diagram that visualises the entire conversation history, giving users complete control over which context is sent to the underlying LLM. This approach prevents context windows from being polluted with irrelevant branches and allows for non-linear, branching conversations – much like a version control system for LLM interactions. Targeting developers, researchers, and power users who work on complex, multi-turn tasks with large language models, Twigg emphasises precision over convenience. Instead of scrolling through endless chat logs, users can see a visual map of their conversation, prune or rearrange branches, and selectively inject context. This makes it ideal for debugging, iterative reasoning, and maintaining coherent context across many exchanges. Under the hood, Twigg acts as a smart middleware between the user and the LLM. It tracks every message as a node in a tree, allows users to fork new branches, merge or delete paths, and then compiles a context window containing only the nodes the user choses. The service integrates with major AI providers and is accessible via a web interface, with potential for future API access. What sets Twigg apart is its deliberate focus on structural context control. While most AI chat tools offer a linear history, Twigg’s tree-based approach mirrors how complex reasoning actually unfolds – non-linearly and with multiple dead ends. For power users who need to revisit, compare, and reuse previous reasoning paths, Twigg offers a level of context fidelity that standard chatbots cannot match.