The Export-Blueprint Journey: Package Your Infrastructure Decisions for Reuse

When I have a working workspace configuration that took weeks to develop, I want to package it as a shareable blueprint, so I can replicate it in other workspaces, share it with colleagues, or version-control it for audit purposes.

Introduction

You’ve spent three weeks building the perfect RAG setup: curated sources from vendor documentation, battle-tested scenarios with production-grade SLOs, and a stack configuration that balances cost and performance. It works. Your team loves it. Now the Singapore team asks: “Can we get a copy of your setup?”

What follows is a familiar disaster. You export your sources as a list of URLs. You screenshot your scenario configuration. You write up the stack settings in a Confluence doc. Three weeks later, the Singapore team is still debugging why their implementation doesn’t work like yours—because they missed a scenario constraint you forgot to document.

The core problem is that infrastructure configurations are complex, interconnected artifacts. Sources inform scenarios. Scenarios constrain stacks. Stack choices affect source retrieval strategies. Sharing any piece in isolation loses the context that makes the whole system work.

How Lattice Helps

The export-blueprint journey transforms a complete workspace—sources, scenarios, and stacks—into a portable YAML file that captures the full configuration. One click packages weeks of work into an artifact that can be imported into another workspace, shared with colleagues, version-controlled in git, or published for community reuse.

The export is selective: you choose which sources to include, which scenarios to share, which stacks to package. The YAML format is deliberately human-readable. You can diff blueprints to see what changed, edit them in a text editor, or review them in a pull request before importing.

The Export-Blueprint Journey in Action

Step 1: Access the Export Modal

From the SourcesPanel header, locate the “Export Workspace as Blueprint” button (file-code icon). Click the button to open the Export Blueprint modal.

Step 2: Define Blueprint Metadata

The metadata fields establish the identity of your blueprint:

Blueprint Name (required): Enter a descriptive name that helps others understand what this blueprint provides.

• Good: “Production RAG for Customer Documentation”
• Avoid: “My Setup” or “Test Blueprint”

Description: Provide context about what this blueprint includes and when to use it.

Version: Track blueprint revisions with semantic versioning. Start with “1.0.0” for initial export.

Category: Select the primary use case: general, production, development, comparison, cost_optimization, rag, or agentic.

Step 3: Select Content to Include

The modal presents expandable sections for each content type:

Sources Section:

• Header shows “Sources (X of Y selected)”
• Master toggle checkbox to select/deselect all
• Each source shows title and type

Scenarios Section:

• Header shows “Scenarios (X of Y selected)”
• Each scenario shows name, category, and key SLOs

Stacks Section:

• Header shows “Stacks (X of Y selected)”
• Each stack shows name, model, and framework

Step 4: Set Visibility Options

Choose who can access this blueprint:

• Private: Only you can see and use this blueprint
• Team: Visible to your organization
• Public: Available to the Lattice community

Step 5: Export and Download

Click “Export Blueprint” to generate the package. The modal shows progress, success confirmation, and a “Download YAML” button.

The downloaded YAML contains:

name: Production RAG for Customer Documentation
description: Battle-tested configuration for customer-facing documentation Q&A
version: "1.0.0"
author: Platform Team
category: production
visibility: team
tags:
  - rag
  - production

sources:
  - title: Anthropic Claude Documentation
    type: url
    url: https://docs.anthropic.com/
    category: vendor

scenarios:
  - name: Production Customer Q&A
    category: inference
    workload_type: rag
    slos:
      latency_p95_ms: 500
      throughput_rps: 100

stacks:
  - name: Claude + LangGraph Production
    model:
      provider: anthropic
      model_id: claude-sonnet-4-5-20250929

Real-World Scenarios

Team Onboarding

A new ML engineer joins the platform team. Instead of spending a week understanding the production setup, they:

1. Import the “Production RAG” blueprint to a personal workspace
2. Explore the sources, scenarios, and stacks
3. Make experimental changes without affecting production
4. Learn the configuration through hands-on exploration

Time to productive: days instead of weeks.

Multi-Region Deployment

The US production team has a working setup. The EU team needs the same configuration adapted for GDPR compliance:

1. US team exports production blueprint
2. EU team imports the blueprint
3. EU team modifies scenarios to add EU regions
4. Both teams maintain consistent base configuration

Architecture Decision Records

Before an architecture review, the team needs to document the current state:

1. Export the production workspace as a blueprint
2. Commit the YAML to the architecture decision record repo
3. Link the blueprint in the ADR document

The blueprint becomes part of the permanent documentation—not just a description of the system, but the actual configuration.

What You’ve Accomplished

By completing the export-blueprint journey, you can now:

• Package complete workspace configurations into shareable YAML files
• Selectively export sources, scenarios, and stacks
• Version-control your AI infrastructure configuration
• Share blueprints with team members or the community

What’s Next

Related capabilities extend the workflow:

• Blueprint Gallery: Browse and apply blueprints from the community
• Apply Blueprint: Import blueprint sources, scenarios, and stacks
• Workflow Templates: Pre-built templates for common patterns

The export-blueprint journey is available in Lattice v0.9.5+. Click the file-code icon in the Sources panel to package your workspace for sharing and reuse.