Learning_AI_using_AI_-_Part_2_-_From_Workflow_Weaver_to_Architect

Tawfik Platform

This course, "Learning AI using AI - Part 2 - From Workflow Weaver to Architect," is a comprehensive builder's sequel that teaches professionals to construct production-grade AI-powered systems. It progresses through three modules, each building on the last: Module 1 (Workflow Weaver) establishes foundational automation pipelines using the cheap-workhorse pattern and three model access modes; Module 2 (Builder) teaches hands-on construction of RAG systems and reasoning-model agents with open-weight models; Module 3 (Architect) covers multi-model orchestration across four axes, fallback reliabi…

Course outline

Module 1 — The Workflow Weaver

This module, "The Workflow Weaver," establishes the foundational architecture for building production-grade AI automation pipelines. It begins by introducing the automation platform landscape (Zapier, Make, n8n) and a trade-off triangle framework for platform selection based on e…

  • Topic 1.1 — Automation Platforms and AI Steps This topic covers the foundational concepts of automation platforms and how artificial intelligence integrates within automated workflows. The first lesson, "The Automation Landscape," introduces three major platforms—Zapier, Make, and n8n—and a trade-off triangle framework for s…
  • Topic 1.2 — The Three Access Modes This topic covers the three primary methods for accessing large language models: first-party APIs, inference providers, and self-hosting. First-party APIs offer the most direct relationship with model creators (Anthropic, OpenAI, Google, DeepSeek) through a credit-card-to-vendor …
  • Topic 1.3 — The Cheap-Workhorse Pattern This topic introduces the "Cheap-Workhorse Pattern," a fundamental cost-optimization architecture for production AI workflows. It establishes that most AI tasks in a typical pipeline are "easy" (classification, extraction, summarization, routing) and can be handled by cheap model…
  • Topic 1.4 — Module 1 Capstone This capstone module synthesizes all Module 1 concepts into a practical, real-world automation project. Students build a complete workflow on platforms like Zapier, Make, or n8n, applying the cheap-workhorse-plus-frontier-polish pattern and making deliberate choices about AI acce…

Module 2 — The Builder

This module, "The Builder," equips learners with the practical skills to construct production-grade AI systems. It begins by establishing a strategic framework for selecting and deploying open-weight models, covering evaluation, licensing, and access-mode decisions across a produ…

  • Topic 2.1 — Open-Weight Models as a Builder's Superpower This topic, "Open-Weight Models as a Builder's Superpower," provides a strategic and practical framework for leveraging open-weight models in AI product development. It begins by defining open weights and distinguishing them from open source, then argues for their strategic advan…
  • Topic 2.2 — Building the Retrieval Pattern Yourself This topic guides learners through building a complete Retrieval-Augmented Generation (RAG) system from first principles to production-grade implementation. It begins with the conceptual foundation of the five-stage RAG pipeline (chunk, embed, store, retrieve, generate), explaini…
  • Topic 2.3 — Reasoning Models as Agent Brains This topic covers reasoning models as the "brain" of AI agents, progressing from foundational concepts to practical implementation. Lesson 2.3.1 introduces reasoning models as language models trained via reinforcement learning to produce visible chain-of-thought, establishing the…
  • Topic 2.4 — Module 2 Capstone This capstone module requires students to integrate at least two of three Module 2 skill areas—open-weight strategy, source-grounded retrieval, and reasoning-agent orchestration—into a single, builder-grade tool. The project spans 8-15 hours over two weeks and demands end-to-end …

Module 3 — The Architect (Weeks 6–8)

This module, spanning weeks 6-8, transitions students from building individual tools to designing robust, production-ready multi-model systems for other engineers. It begins by introducing the foundational four-axis routing framework (capability, cost, latency, access mode) and p…

  • Topic 3.1 — Routing Across Four Axes This topic marks the transition from building tools to designing systems for other engineers, introducing the core framework for multi-model orchestration. It establishes that every routing decision simultaneously resolves four constraints—capability, cost, latency, and access mo…
  • Topic 3.2 — Fallback, Reliability, and Evaluation This topic covers the full production reliability toolkit for LLM routing systems, moving from reactive failure handling to proactive quality assurance. Lesson 3.2.1 introduces fallback chains, circuit breakers, and graceful degradation as resilience mechanisms against provider d…
  • Topic 3.3 — Governance, Sovereignty, and Provider Diversification This topic addresses the critical non-functional requirements for production AI systems: governance, sovereignty, and provider diversification. Lesson 3.3.1 covers strategic vendor risk management, detailing four categories of concentration risk and the solution of multi-provider…
  • Topic 3.4 — Module 3 Capstone This capstone combines the build and reflection phases of Module 3. Lesson 3.4.1 requires students to integrate previously built components into a working multi-model production system, meeting six hard requirements including provider diversity, access mode variety, cross-provide…

What you get

AI tutor in context

Ask about any lesson; the tutor answers from this course's own material.

Hint ladder

Step-by-step hints that guide you to the answer instead of giving it away.

Quizzes with scoring

Check your understanding and surface weak spots as you go.

Debate mode

Argue a concept with the tutor to deepen and test your grasp.

XP, streaks & badges

Steady study turns into visible progress across the course.

Sign in to start this course →