Why This Playbook?
Generative AI is often adopted quickly, but production usefulness depends on clear output contracts and safe fallbacks. This playbook shows a practical, mock-first implementation that teams can run locally in minutes.
Repos:
- Generative: github.com/amiya-pattnaik/generativeAI-engineering-playbook
- Agentic: github.com/amiya-pattnaik/agentic-engineering-playbook
- RAG: github.com/amiya-pattnaik/rag-engineering-playbook
Concept Primer: What Is Generative AI?
Generative AI uses LLMs to create new content from prompts (for example drafts, summaries, checklists, test ideas). In this repository, the model generates structured artifacts from a task + context input.
Broader GenAI Use Cases
- Requirement and user-story drafting.
- PR/change summaries and release notes.
- Incident/postmortem first drafts.
- Knowledge article/support response drafting.
- Test strategy and checklist generation.
Demo scope in this repo:
- For repeatability, this demo focuses on test case generation from
task + context.
Concept Comparison (GenAI vs Agentic vs RAG)
User Need
|
+--> Fast content draft from prompt/context
| -> Choose GENERATIVE AI
|
+--> Multi-step planning + tool orchestration
| -> Choose AGENTIC AI
|
+--> Answers grounded in source documents with citations
-> Choose RAG
What It Demonstrates
- Structured prompt contract with JSON output.
- Guardrails for parsing/normalizing model output.
- Mock-first mode for offline demo reliability.
- Provider mode using OpenAI with no UI changes.
- Scenario runner for repeatable runs and report artifacts.
Flow
- User submits
task + contextfrom UI. - API validates request and builds structured prompt.
- Model selector chooses mock or provider.
- Model returns JSON output.
- Server validates/parses response and applies fallback if needed.
- UI renders normalized structured result.
ASCII Diagram
User (UI)
|
v
POST /api/generate -> Validate Input -> Build Prompt
|
v
Model Selector (Mock/OpenAI)
|
v
JSON Completion
|
v
Parse/Guard/Normalize Output
|
v
UI Response
Provider Support
- OpenAI is integrated out-of-the-box.
- Other providers (Gemini, Claude, etc.) can be added via provider adapters in
demo-app/src/providers/and provider-selection logic insrc/services/generator.js.
Quickstart
git clone https://github.com/amiya-pattnaik/generativeAI-engineering-playbook.git
cd generativeAI-engineering-playbook/demo-app
cp .env.example .env
npm install
npm run dev
# open http://localhost:3000
Use OpenAI:
- Set
OPENAI_API_KEY(optionalOPENAI_MODEL) in.env.
Run and Evaluate
npm run demo:scenarios
npm run demo:scenario -- scenarios/login-mfa.json
Outputs are written to demo-app/reports/ (JSON + Markdown).
Closing Thought
Generative AI is strongest when treated as an engineering interface with contracts, validation, and clear fallback behavior, not as an unconstrained text box.