100 Solution Plays — A Playbook for Every AI Architecture
When a team asks "how do I build a RAG pipeline?", the answer shouldn't be "it depends" followed by weeks of research. It should be: Play 01 — Enterprise RAG.
That's the vision behind FrootAI's 100 Solution Plays — a curated catalog of production-ready AI architectures, each complete with:
- Agents — Builder, Reviewer, and Tuner for every play - Skills — Step-by-step procedures for implementation - Infrastructure — Azure Bicep templates (where applicable) - Evaluation — Quality thresholds and test pipelines - WAF Alignment — All 6 Well-Architected Framework pillars
The Categories
Foundation (Plays 01-10)
The building blocks. Start here.| Play | Name | Key Services | |------|------|-------------| | 01 | Enterprise RAG | Azure OpenAI + AI Search + Cosmos DB | | 02 | AI Landing Zone | Hub-spoke networking, RBAC, monitoring | | 03 | Deterministic Agent | Zero-temperature, structured output | | 04 | Call Center Voice AI | STT → LLM → TTS streaming | | 05 | IT Ticket Resolution | ServiceNow + Azure OpenAI |
Advanced (Plays 11-20)
For teams ready to scale.| Play | Name | Key Services | |------|------|-------------| | 11 | AI Landing Zone Advanced | Multi-region, private endpoints | | 13 | Fine-Tuning Workflow | LoRA, QLoRA, JSONL data prep | | 14 | Cost-Optimized AI Gateway | Model routing, token budgets | | 17 | AI Observability | Custom metrics, drift detection |
Cutting Edge (Plays 21-30)
The frontier.| Play | Name | Key Services | |------|------|-------------| | 21 | Agentic RAG | Agents that decide when/how to search | | 22 | Swarm Orchestration | Multi-agent teams with supervisors | | 23 | Browser Agent | Playwright + LLM web automation | | 28 | GraphRAG | Entity extraction, knowledge graphs |
Industry (Plays 31-100)
Domain-specific patterns for healthcare, finance, legal, manufacturing, and more.How to Use a Play
Option 1: Copilot Agent
``
@fai-play-01-builder Set up an Enterprise RAG pipeline for our HR knowledge base
`Option 2: MCP Server
bash
npx frootai-mcp@latest
Then in Copilot: "Get play detail for play 01"
`Option 3: Browse Online
Visit frootai.dev/solution-plays for the full catalog with architecture diagrams.Play Structure
Every play follows the golden template:
solution-play-NN/
├── agent.md ← Root orchestrator
├── .github/
│ ├── copilot-instructions.md ← Domain knowledge (<150 lines)
│ ├── agents/ ← Builder, Reviewer, Tuner
│ ├── skills/ ← Implementation procedures
│ └── hooks/ ← Guardrails
├── config/ ← Tunable AI parameters
├── infra/ ← Azure Bicep (when applicable)
└── evaluation/ ← Quality pipeline
``Community Contributions
We welcome new plays! If you've built an AI architecture that others could benefit from, submit it to the catalog. See our contribution guide for details.
Browse the full catalog at frootai.dev/solution-plays