Skip to content

Blueprints

Blueprints are curated knowledge bundles that give you a head start on AI infrastructure research. Instead of manually gathering documentation, apply a blueprint to instantly populate your workspace with relevant sources, scenarios, and stacks.

Curated Sources

3-6 official documentation links, pricing pages, and technical guides from vendors

Pre-configured Scenarios

1-2 workload configurations with realistic SLOs and constraints

Recommended Stacks

1-2 infrastructure configurations optimized for common use cases

Suggested Prompts

Research questions tailored to the blueprint’s focus

Official documentation from major AI providers:

VendorFocus Areas
AnthropicClaude models, prompt engineering, safety
OpenAIGPT-4, embeddings, fine-tuning
AWSBedrock, SageMaker, infrastructure
GoogleVertex AI, Gemini, TPU optimization
NVIDIANIM, TensorRT, GPU optimization
CohereCommand, embeddings, RAG
MistralOpen-weight models, deployment

Curated for specific workloads:

Use CaseSources Include
RAG PipelineVector DBs, chunking strategies, retrieval patterns
Fine-tuningDataset preparation, training best practices
Cost OptimizationPricing comparisons, caching strategies
Latency OptimizationBenchmark data, streaming, edge deployment
Enterprise SecurityCompliance docs, security best practices
  1. Navigate to Blueprints in the main navigation
  2. Browse by vendor or search for specific topics
  3. Click a blueprint to view its contents
  4. Click Apply to Workspace to import everything
  5. Wait for sources to index (usually 30-60 seconds)
  6. Start researching with pre-populated context
name: Anthropic Claude
description: Complete Claude documentation and best practices
sources:
- url: https://docs.anthropic.com/claude/docs/models-overview
type: documentation
- url: https://docs.anthropic.com/claude/docs/prompt-engineering
type: documentation
- url: https://www.anthropic.com/pricing
type: pricing
- url: https://docs.anthropic.com/claude/docs/rate-limits
type: documentation
scenarios:
- name: High-Volume Claude Chat
workload_type: chat
traffic_profile: high_volume
slo_requirements:
p95_latency_ms: 500
throughput_rps: 500
stacks:
- name: Claude Sonnet Production
model:
provider: anthropic
model_id: claude-sonnet-4-20250514
name: RAG Pipeline
description: Build production RAG systems
sources:
- url: https://docs.pinecone.io/docs/overview
type: documentation
- url: https://python.langchain.com/docs/modules/data_connection/
type: documentation
- url: https://docs.anthropic.com/claude/docs/retrieval-augmented-generation
type: documentation
- url: https://www.llamaindex.ai/blog/evaluating-the-ideal-chunk-size
type: blog
scenarios:
- name: Enterprise RAG
workload_type: rag
traffic_profile: medium_volume
slo_requirements:
p95_latency_ms: 2000
throughput_rps: 100

Documentation changes over time. Refresh blueprints to:

  • Re-fetch source content
  • Update to latest documentation
  • Maintain accuracy
  1. View the applied blueprint in your workspace
  2. Click Refresh on individual sources
  3. Or click Refresh All to update everything

Want a blueprint for a specific vendor or use case? Request it on GitHub.

GET /api/blueprints

Returns all available blueprints.

GET /api/blueprints/{blueprint_id}

Returns blueprint details including sources, scenarios, and stacks.

POST /api/workspaces/{workspace_id}/blueprints/{blueprint_id}/apply

Imports all blueprint contents into the specified workspace.

GET /api/blueprints?query=anthropic&category=vendor

Filter blueprints by search query and category.