From one AI capability to a full business process
AI Foundations 2: Business Processes builds upon concepts and tools, focusing on how data, systems, and human judgment can collaborate. The quality of a single response is only one aspect; the reliability of sources, how tools are handed off, who has authority to execute, and how results are verified are equally critical to ensuring workflow effectiveness.
Recommended reading order
Read first From rule automation to intelligent workflows , understand triggers, conditions, actions, and feedback, then compare no-code with AI-assisted coding. Next read Data pipelines and Service selection : workflow capabilities are influenced by data quality, service dependencies, integration methods, and maintenance conditions.
Then explore Business Agents objectives, planning, tools, memory, and handover, and use a trustworthy AI framework to monitor and assign responsibility. Only after these conditions are established should you proceed to connecting business knowledge by meaning.
Semantic retrieval, RAG, and MCP integration
Vector databases and semantic search Explain how data is represented and retrieved; RAG Describe how retrieved data is incorporated into responses; MCP and API Explain how applications connect to external tools and systems. Retrieval, generation, and execution each have distinct responsibilities; none can automatically guarantee the correctness of the others.
Then compare intelligence with agency, discussing action outcomes and human accountability; next introduce MoE mixed expert models, differentiating between sparse experts within a model, multi-model selection, and collaboration among multiple agents. A comprehensive exercise integrates prior knowledge into a business workflow with evaluation metrics and governance arrangements.
Comprehensive explanation and practical demonstration
The content is divided into two parts based on real-world stages: Workflow mapping and organizational knowledge First explain workflows, implicit rules, ontologies, and operational frameworks; Tool integration and invoice processing Agent Continue with API, MCP, image, and 3 D examples, and finally present a complete invoice project workflow, including instructions, execution, queries, failures, and corrections.
Keep both lecture notes and spoken text separate for comparison between concepts and implementation. Examples do not imply that this site provides these automation services, nor do they consider tool prompts indicating 'completion' as evidence of correct results.
How to read the full text on trustworthy AI
Understand Trusted AI Primer Afterward, you can read sequentially the 'Ethical Guidelines for Trustworthy AI' including its framework 、 principles 、 requirements and methods 、 evaluation checklist and application examples and terminology . Together, these five texts cover the substantive content of the 2019 version, with attention paid to historical policies and predictions.
Another document full text of the Hiroshima AI Behavior Guidelines focuses on eleven actions for organizations developing advanced systems, followed by a return to reading reflection . Both documents are included in the same reading directory for this course, with English translations providing version information, original texts, licensing details, and modification notes.
Document status and gaps
This directory includes 21 reading materials: 15 main readings from 14 available units, with one unit split into two pieces; additionally, there are 6 independent reference translations, including the substantive content of the two documents. The ethics definition pairing exercise lacks actual practice questions and no blank exercises or guessed questions were created. The two full reference texts have now been added; the papers on intelligence and agency currently only provide an outline, supporting text, and query links, without full translation. Private responses are not publicly available. Subsequent live sessions not yet obtained are not counted as translated.
The same eight video transcripts from the first course share complete translations, directly jumping to the relevant sections; the new course's unique main texts, videos, and examples remain separately preserved. When reviewing concepts, you can return to AI Foundations 1: Concepts and Tools .