Differentiate AI roles - You will clearly distinguish a chatbot, an AI assistant, and an AI agent, so you can match the right tool to the task and avoid deploying solutions that do not fit your process.
Design AI workflows - You will turn a real business process into a single-step or multi-step AI workflow, from inputs and decision points to output validation and a usable result for your team.
Organize knowledge in Claude - You will learn how to structure Claude Projects, manage documents, control context, and build project memory so your agent can work from current and reliable company knowledge.
Build no-code tools - You will create practical solutions in Claude Artifacts, including AI forms, dashboards, calculators, and document generators that can support your team's daily work right away.
Write stronger prompts - You will learn how to define agent roles, goals, constraints, and operating instructions to run multi-step processes, automate analysis, and gain tighter control over model behavior.
Connect Claude to tools - You will see how to use MCP to securely connect Claude with company documents, knowledge repositories, SharePoint, or Google Drive and give agents the context they need.
Automate document analysis - You will practice working with PDFs, contracts, reports, and technical documentation, from data extraction and classification to recommendations and a final report ready to use.
Deploy AI safely - You will prepare an implementation plan for an AI agent that covers data protection, access rights, auditing, monitoring, and human-in-the-loop controls for safe automation.
AI agents differ from chatbots and assistants because they can take actions within a defined workflow built on language models. The opening section establishes these distinctions, shows business use cases, and situates Claude Desktop, Projects, Artifacts, and API within one platform for process automation.
Designing an AI workflow begins with mapping business processes and identifying tasks suitable for automation. From there, the section develops workflow architecture covering inputs, information processing, decision points, result generation, and validation, including both single-stage and multi-stage process structures.
Within Claude Projects, the agent works in a structured environment built from documents, repositories, context, and project memory. The section explains how to organize procedures, instructions, and project materials into a usable knowledge base, then configure specialized agents for analysis, documentation, projects, and team support.
With Claude Artifacts, participants build AI tools without programming, including forms, dashboards, document generators, calculators, and reporting components. Attention is given to iterative development: testing functionality, refining behavior, extending workflows, and managing versions as the solution evolves.
System prompts shape how an agent operates by defining its role, goals, constraints, and behavioral rules. The section shows how to build multi-step processes with Chain of Tasks, automate analysis and decision flows, and prepare reusable templates for reporting, analytical, sales, and administrative work.
MCP defines how Claude connects to external systems while sharing data with AI models in a controlled way. This part explains the protocol architecture, integration with repositories such as SharePoint and Google Drive, and the setup of an agent workspace with managed sources, updates, and permissions.
Document-heavy processes can be automated through agents that read PDFs, contracts, reports, and technical documentation. The focus is on multi-stage analysis: extracting data, classifying information, generating recommendations, and preparing final reports with legal, financial, HR, or project-specific context.
Team collaboration often breaks down around follow-ups, summaries, and scattered project knowledge. Here, agents are applied to task management, communication, document and presentation drafting, and administrative support, with Human-in-the-loop practices for delegation, supervision, and result validation.
Coordinating multiple agents starts with role division, specialization, and structured information exchange across a shared process. Participants examine orchestration patterns such as task queueing, decision paths, and escalation, using scenarios from reporting, customer service, project work, and document management.
Secure deployment of AI agents requires clear rules for data protection, confidential information handling, access permissions, and risk management. The section also covers governance mechanisms, then moves into a workshop where participants design, test, validate, and plan implementation of a complete agent-based solution.
Computer basics - You should be comfortable working on a computer and in a web browser, using online documents, and moving efficiently between applications used in your daily work.
Process awareness - You should understand at least one process from your role or company so you can map tasks, identify stages, and assess what can realistically be automated.
Document handling - You should know how to work with documents such as procedures, instructions, reports, or contracts, because you will analyze and organize this kind of material.
Basic AI usage - You should have basic experience using AI tools and writing simple prompts, so you can move more easily into designing agents and building more advanced workflows.
