Generate Architecture

The MCP server includes a generate_workspace prompt that primes the agent with the workspace schema — entity types, required fields, and the shape of each relationship. Always run validate after generation to catch mismatches (enum values in particular are a known source of errors: the prompt and the canonical schema are not always perfectly aligned, and the validator is the authoritative check).

How It Works

Feed your agent any source material and ask it to create the architecture:

“Here’s our IT landscape wiki page. Create a complete Albumi workspace — applications, integrations, data objects, and business capabilities.”

“I’m attaching a spreadsheet with our application list. Import them all into Albumi.”

“We have 3 main systems: Salesforce (CRM), SAP (ERP), and a custom data warehouse. Salesforce sends customer data to SAP via REST API daily. Create the full architecture.”

The agent generates a workspace JSON with proper UUIDs, referential integrity, lifecycle metadata, and all required fields.

What Gets Generated

The agent can create all entity types:

• Organizations — company structure
• Business Capabilities — what the business does
• Data Objects — what data exists, classification, PII flags
• IT Components — databases, platforms, frameworks
• Applications — with lifecycle, TIME classification, criticality, ownership
• Integrations — source, target, protocol, data objects carried
• Initiatives — strategic projects with affected applications

After Generation

The agent automatically:

1. Validates the workspace — schema compliance, referential integrity, circular reference detection
2. Audits — runs 70+ quality and compliance checks
3. Pushes as an Architecture Change Request — you review and approve in the UI

Nothing changes in your production workspace without your explicit approval.

Tips

• The more detail in your source documents, the better the result
• You can generate incrementally — start with applications, add integrations later
• Ask the agent to fill in missing fields: “Infer business criticality from the descriptions”
• After generation, ask: “Run a full audit and fix any issues”