Language and reasoning
Prompting, structured outputs, schema validation, reasoning workflows, and code generation.
Open-source research software · Version 1.0.0
AI APIs
in Practice
A visual and executable AI API playbook for multimodal, agentic, production-ready, and commercially realistic applications—created by Faramarz Kowsari.
Capability map
From API foundations to production systems
The repository is organized around durable engineering capabilities rather than short-lived model names, making it useful for learning, experimentation, architecture design, and commercial prototyping.
Agents and tools
Function calling, tool use, agent orchestration, computer interaction, search, and MCP patterns.
RAG and knowledge
Embeddings, chunking, retrieval, reranking, vector stores, and RAG evaluation.
Multimodal AI
Vision, OCR, image generation, speech, realtime voice, video, and media pipelines.
Production engineering
Security, privacy, retries, observability, caching, routing, budgets, and cost controls.
Commercial blueprints
Product concepts based on measurable outcomes, unit economics, safeguards, and human review.
Explore the project
Executable examples, documentation, and citation
Move directly from the conceptual map to runnable code, architecture guidance, provider coverage, and preserved research metadata.
Executable Colab showcase
Run the project without paid API credentials
Source-code examples
Browse focused provider and capability examples
Architecture guide
Read the production-oriented design principles
Provider matrix
Review provider coverage and verification status
Commercial blueprints
Explore outcome-oriented AI product concepts
Permanent scientific record
Cite version 1.0.0 through its Zenodo DOI
