illumi

illumi is an AI visual workspace where notes, research, prompts and outputs live as connected cards on an infinite canvas. Users run multiple AI models from selected context, compare answers, and collaborate with teammates before shaping a final deliverable.

illumi is a visual workspace built for complex, thinking-heavy work that does not fit inside a single AI chat. Instead of a linear conversation, it offers an infinite canvas where ideas, notes, files, research, prompts and AI outputs sit as connected cards. The aim is to keep the reasoning behind a piece of work visible, so each new step can build on what came before rather than starting from a blank prompt. In practice, a user brings in rough inputs, groups related items into sections, and draws connections between the pieces that belong together. Prompts are then run from specific inputs, which means the model only sees the context the user selected. Noisy or outdated inputs can be removed and prompts edited as the work changes, which keeps outputs from being shaped by everything ever added. The same context can be sent to several AI models (the site cites support for more than 30), and the responses can be compared side by side to decide which direction deserves more effort. AI outputs can then be edited and refined in the same space and shaped into a deliverable such as a deck, plan, report or proposal. Collaboration is part of the design. Teammates, clients or workshop participants can contribute notes and review the inputs, prompts and outputs behind a result, so feedback lands while the thinking is still open. The homepage points to research synthesis, strategy work, go-to-market planning, client proposals and workshop follow-through as the strongest fits, and names consultants, founders, researchers and strategists as typical individual users. Among alternatives, illumi sits between whiteboard-style canvases, note-taking tools and multi-model chat front ends. User testimonials describe it as something like Notion combined with AI generation tools. People who only need quick answers from one model will find a plain chat interface simpler, while those juggling several sources and iterations get a structured place to keep them together. A free start is offered, and a demo can be booked for team use.