From the isolated prompt to a collective asset
AI answers. Marylink makes your whole organization answer. This carousel follows the journey of a practicePracticeA unit of know-how captured in Marylink: not a document but an executable structure (content, prompt, rules, style).: from the throwaway prompt, forgotten the moment it has served, to the living building block, governed, shared, capitalized, that becomes a collective asset. Ten boards to understand how.
The isolated prompt
Every AI conversation starts from scratch. One use, then it's forgotten. Another use, then forgotten again. Between two exchanges, no memory is carried over: the practice vanishes as fast as it is created.
You installed the AI. You did not install the memory.
Atomic AI
The monolithic prompt breaks into bricks. Atomization separates three distinct elements: Content (the domain knowledge), the Prompt (the intent to act) and Style (the modulation).
Three bricks. Living. Recombinable.
Spaces and roles
Each brick finds its place depending on who uses it. SpacesSpaceA workspace by domain or topic where a team publishes, shares and governs its practices., Personal, Shared, Official, Domain, Project, say where a brick lives. Roles, from Member to Admin, say who can do what. The Personal space stays open to everyone; the Official, Domain and Project spaces are reserved for the roles that produce.
The space says where. The role says who. GovernanceGovernanceThe roles, validation steps and reviews that ensure the quality of shared practices. writes itself.
Living components
A brick is not a file: it is a publicationPublicationA practice published in a space: versioned, reviewed and subject to validation steps.. The component lives in a cycle, it is edited, commented, reviewed, versioned. It carries a history, feedback, a rating, versions.
Not a dead wiki. A social component, living and governed.
The MCP channel
You reuseReuseThe same practice serving many times, across many spaces, the key measure of the Practice Graph's value. the agentic AI of the chatbots you already have, Claude, ChatGPT, Gemini, Mistral, through their subsidized subscription. MCPMCPModel Context Protocol: the open standard that connects an AI assistant (Claude, etc.) directly to your Marylink space. (Model Context Protocol) connects them to your network: they reason, your bricks supply.
Their brain, your memory. Their tokens, your asset.
Your network
Your bricks form a Practice GraphPractice GraphThe living architecture that links your practices, concepts, roles and spaces, executable by your teams and by AI.: Contents, Prompts, Styles and composition Tools, linked together. This graph is your property and stays provider-independent, OpenAI, Google, Anthropic remain interchangeable, never owners of your knowledge.
The LLMs come and go. Your Practice Graph stays.
The executable graph
The Practice Graph is not just something to browse: it is executable. AI is active within it, it creates, activates, composes and orchestrates the bricks by following the graph's paths.
A living graph. An agent that knows which way to go.
The leverage effect
Few nodes, many tools. 6 well-governed bricks recombine into 30 tools: on average, one brick is used in five tools. You maintain the source, not the multiplication.
You don't maintain thirty tools. You maintain six bricks.
Your turn
Your first bricks take an hour, in four moves: 1. Identify your key practices, 2. Atomize them into bricks, 3. Connect via MCP, 4. CapitalizeCapitalizationTurning know-how produced along the way into a reusable asset, instead of losing it after each use. in your network. Your teams' know-how deserves better than a prompt that dies, where will you start?