Features

Allow users to publish a Membase wiki, or a specific folder within a wiki, publicly so others can add it to their own account. Use cases: An educator shares class notes, slides, or reference material. A creator shares a curated knowledge base instead of a static PDF. A subject-matter expert publishes an industry knowledge base. A team shares onboarding context for new collaborators. Suggested sharing model: Users share specific folders rather than their entire wiki. Two possible modes: One-time snapshot download - easier to ship first, lower risk, easier to vet for safety. Followed knowledge base - the imported wiki stays updated in the user's account. More powerful long term but requires stronger safety controls. Desired user flow: From inside Membase, a user clicks a publish/share button, lands on a public-facing page, and lets others add the knowledge base to their own account. If the user is already signed in, clicking "Add" automatically imports it. Safety requirements (before any wiki goes public): Membase should scan published wikis for prompt injection attempts and dangerous or malicious content before they become public. Imported wikis should be clearly labeled as external in the user's account. Starting with snapshot imports before allowing live-updating followed wikis reduces risk while validating demand. Clear separation between personal memory and imported content: When users import a public knowledge base, Membase should clearly distinguish it from personal memory and user-created wiki content. Imported wikis should appear as clearly marked external folders. When an AI uses imported wiki content, it should understand the content is external reference material, not personal memory. Recommended approach: source-tagged hierarchical retrieval, where personal memory gets priority and retrieval falls back to imported wikis only when needed. Long-term evolution - Marketplace for Knowledge Bases: Over time, public knowledge bases could evolve into a marketplace where creators publish high-quality curated wikis. Creators could charge for access, with Membase taking a commission. This creates a new revenue stream, incentivizes quality content, and turns every shared wiki into a distribution loop that brings new users into Membase. Why it matters: Public knowledge bases could become a distribution channel for Membase. Every shared wiki becomes a way for new users to discover the product through useful content. With the right safety controls and clear separation between personal and external context, this feature cluster could meaningfully expand what Membase is useful for.

User problem: Collaborating with others using AI currently requires someone to manually create and maintain a separate context document. This adds friction, especially for collaborators who are not technical or do not already know how to set up AI memory tools. Desired behavior: Create Shared Spaces inside Membase: collaborative workspaces that combine wiki content and memories, allowing multiple people to access the same project context. One person creates a shared workspace, invites collaborators, and gives them immediate access to the relevant project context. Suggested onboarding flow: Email invite -> click link -> create account -> connect AI client -> enter shared workspace with project context already available Key requirements: An owner/admin manages access and can delete entries. Users manually decide what enters the shared space. Personal memories never bleed into shared spaces. Support both shared wiki folders and full shared workspaces. Enterprise angle: Shared Spaces could support enterprise adoption if paired with proper compliance and controls such as audit trails, role-based access, compliance readiness, and per-seat or per-workspace pricing. Why it matters: Shared Spaces would make Membase useful not only for individual memory, but also for collaborative projects, teams, classrooms, and companies. It also creates a natural growth loop: one user invites collaborators, those collaborators experience value immediately, and some later adopt Membase personally.

Current behavior: The AI usually needs the user to explicitly say something like "Search Membase" to retrieve wiki content. Alternatively, the AI may pick up on memories automatically, but there is no reliable way to define when a specific wiki folder should be checked. User problem: Users cannot set persistent retrieval instructions such as: "Whenever I ask about university classes, check the syllabus folder." As a result, they have to repeat the same instruction across conversations. Desired behavior: Add a "when to use this" or trigger/context field at the folder or file level, similar to the trigger/description field used by skills. This would allow users to define contextual usage instructions once. Examples: Wiki folder: "University syllabi" | Trigger: "Use whenever the user asks about courses, exams, assignments, readings, or university deadlines." Wiki folder: "Startup docs" | Trigger: "Use whenever the user asks about company strategy, product decisions, investor updates, or roadmap planning." Why it matters: This would make wikis feel more autonomous, persistent, and context-aware. Users could set retrieval behavior once and rely on the AI to use the right knowledge source at the right time, without repeating instructions every conversation.