Anuma

Anuma is an AI chat bot tool. A multi-model AI chat app built around one encrypted memory that follows you across ChatGPT, Claude and Gemini, with image, audio and video creation, no-code app building, and a Nearby map that suggests people close by who share your interests.

Anuma is a chat platform that places a single encrypted memory at the centre of the AI experience. Instead of keeping separate accounts with ChatGPT, Claude and Gemini, users work inside one app where the same set of remembered facts follows them from model to model and from device to device. The memory is not a passive log: users choose what gets stored, and can view, edit or delete individual memories at any time. Context such as a training plan, a dietary change or an ongoing project is captured once and reused across every conversation, so switching models no longer means repeating yourself. Beyond conversation, Anuma is positioned as a workspace for making things. Users can generate images, audio and video from a prompt with a chosen style, and can assemble custom tools, workflows and small apps by describing them in plain language, with any available model powering the result. Chats, creation and builds all draw on the same stored context, which is the main structural difference from chatting with a single-vendor assistant. The most distinctive feature is Nearby, a map-based discovery layer that reads the interests held in memory and surfaces people close by with overlapping profiles. Anuma is explicit that this is not a dating product: the framing is shared hobbies, local activities and meetups. Menus for dinner parties, ramen recommendations and race training plans all become match signals, and users see a proximity distance along with a stated reason for the match before deciding to interact. Privacy is treated as a configurable setting rather than a fixed promise. Conversations are encrypted in transit and at rest, backup is optional, and Anuma states that private memory is not sold or used to train models. Closed-source providers still receive the messages needed to answer a prompt, under their own retention rules. The product is aimed at people who already use several assistants and want one memory layer, and at those curious about interest-based local discovery. It runs on the web plus iOS and Android.