Course code: PT1008
Apply advanced prompt-engineering techniques for reliable generative-AI output across business and engineering tasks. GenAI fundamentals cover tokenization, embeddings, decoding, and the model settings that shape output quality. Prompting techniques cover zero-shot, few-shot, and dynamic example selection, alongside the practical limits of N-shot. Chain-of-thought, step-based, meta, and reflective prompting cover CoT variants from zero-shot through step-back, prompt chaining, task decomposition, Reflexion-style self-correction loops, and emerging techniques including Logic-of-Thought and Narrative-of-Thought. Prompting agentic AI and advanced techniques apply the patterns to tool use, memory integration, goal-oriented prompts, ensembling, code prompting, self-consistency, self-criticism, and plan-and-solve. Hands-on labs produce a personal library of prompt patterns, an agentic prompt set with tool-use scaffolding, and a self-correcting chain-of-thought workflow. The course is designed for anyone using generative AI day-to-day, including business users, analysts, data scientists, and developers with introductory GenAI exposure.
Course code: PT1000
Apply generative AI to daily workplace tasks without becoming an AI specialist. Generative AI foundations cover what GenAI is and how it differs from traditional AI, tokenization, embeddings, decoding, foundation model architectures, and the model settings such as temperature, top-p, and max tokens that shape output quality. Prompt engineering covers the CLEAR framework and RCTO/RTF structures, the adaptive iteration loop, and workflow decomposition with human checkpoints between stages. Use cases and ethics cover real-world applications, the ethical considerations around responsible use, and the regulatory and governance issues organizations now face. Hands-on no-code labs produce a personal prompt library, a workplace-task workflow, and case-study analyses across cross-industry examples. The course is designed as onboarding for working professionals across business, technical, and operations roles adopting AI tools.
Course code: PT1002
Apply generative AI to daily workplace tasks without becoming an AI specialist. Generative AI foundations cover what GenAI is and how it differs from traditional AI, tokenization, embeddings, decoding, foundation model architectures, and the model settings such as temperature, top-p, and max tokens that shape output quality. Prompt engineering covers the CLEAR framework and RCTO/RTF structures, the adaptive iteration loop, and workflow decomposition with human checkpoints between stages. Use cases and ethics cover real-world applications, the ethical considerations around responsible use, and the regulatory and governance issues organizations now face. Hands-on no-code labs produce a personal prompt library, a workplace-task workflow, and case-study analyses across cross-industry examples. The course is designed as onboarding for working professionals across business, technical, and operations roles adopting AI tools.
Course code: PT1003
Run Claude Code as a capable AI coding agent rather than an autocomplete tool. AI coding agent architecture covers context windows, token limits, the plan-execute-iterate cycle, autonomous file and terminal operations, and the permission boundaries that keep agentic operation safe. Context engineering covers prompting strategies, file and codebase reference systems, extended-thinking budgets, and cross-session memory that shapes agent behaviour over a project’s lifetime. Configuration standards, the Model Context Protocol, and practical workflows cover CLAUDE.md, AGENTS.md, exclusion strategies, MCP hosts and servers, permission management, code generation, review, testing, debugging, and refactoring. Claude Code product features cover terminal-first and IDE operation, checkpoints and background tasks, Skills, hooks, custom slash commands, and headless operation via the Claude Agent SDK. Hands-on labs produce a configured environment, a project-level CLAUDE.md, a reusable Skill, at least one MCP integration, and a working agent-driven Git workflow. The course is designed for software developers, data scientists, and anyone integrating an AI coding agent into daily team workflows.
Course code: PT1004
Run GitHub Copilot CLI as a capable AI coding agent rather than an autocomplete tool. AI coding agent architecture covers context windows, token limits, the plan-execute-iterate cycle, autonomous file and terminal operations, and the permission boundaries that keep agentic operation safe. Context engineering covers prompting strategies, file and codebase reference systems, extended-thinking budgets, and cross-session memory that shapes agent behaviour over a project’s lifetime. Configuration standards, the Model Context Protocol, and practical workflows cover AGENTS.md, repository custom instructions, exclusion strategies, MCP hosts and servers, permission management, code generation, review, testing, debugging, and refactoring. GitHub Copilot CLI product features cover terminal-first operation, hooks, custom agents, built-in specialized agents, and native GitHub issue and PR integration. Hands-on labs produce a configured environment, a project-level AGENTS.md, at least one MCP integration, and a working agent-driven Git workflow. The course is designed for software developers, data scientists, and anyone integrating an AI coding agent into daily team workflows.
Course code: PT1005
Build production generative AI applications with hosted LLMs, LangChain, and agentic patterns. The course opens with an optional GenAI foundations recap, then covers prompt engineering with CLEAR and RCTO, hosted-API chatbot construction, LangChain primitives, RAG, and evaluation through metrics, custom criteria, and tools including DeepEval and G-Eval. Production concerns cover deployment and scaling tradeoffs across cloud and on-premise, latency and throughput management, monitoring with alerts and dashboards, and debugging strategies that survive non-determinism. Agentic AI with LangGraph covers agentic patterns, multi-agent workflows, and error handling. Hands-on labs produce a chatbot, an evaluated LangChain pipeline, a deployed and monitored GenAI service, and a working agentic router. The course is designed for software developers and data scientists with Python and introductory generative-AI experience.
Course code: PT1006
Build no-code agentic AI workflows for business automation and collaboration through n8n, applied throughout to health insurance scenarios such as member service, claims, and prior authorization. Agentic AI foundations cover what agentic AI is, how it differs from other AI techniques, and the collaboration and automation use cases that show up in a health plan. The components of an AI agent cover foundation models, model selection, working and episodic memory, and the OODA decision pattern. Tool use and MCP servers cover observation versus action tools, direct and indirect tool-use methods, and integrating MCP servers for shared capabilities. Building agentic applications, designing collaborative agents, and the ethics and security of agentic AI cover task decomposition, multi-agent systems, automation versus human collaboration, fairness, transparency, accountability, security threats, and excessive-agency risks. Hands-on labs on the n8n platform produce working health insurance agents, human-in-the-loop workflows, and an excessive-agency risk audit. The course is designed for business users learning to apply agentic AI at a health insurance organization.
Course code: PT1007
This course gives senior directors a leader's-level view of AI strategy, risk, and enterprise readiness for a health insurance organization. Participants learn how generative and agentic AI work, plus the vocabulary used in claims, underwriting, and utilization management. They learn to spot high-value use cases — claims, prior authorization, fraud, member engagement — and choose KPIs that avoid pilot purgatory, like claims cycle time and medical loss ratio. Risk topics include health-insurance-specific AI risks, Responsible AI principles, the NAIC Model Bulletin, and HIPAA-driven vendor requirements. Enterprise readiness covers adoption roadmaps and talent strategy across actuarial, clinical, claims, and compliance teams. Labs produce a KPI grid, a risk map, an investment portfolio, and a personal adoption roadmap.
Course code: PT1009
This Azure Databricks training teaches learners proven, real-world techniques to leverage the power of cloud data engineering and analytics on the Microsoft Azure platform. Learners explore fundamental Big Data principles, the practical applications of Apache Spark, and hands-on utilization of Azure Databricks for scalable data engineering and analysis. This comprehensive, hands-on course focuses on practical skills, giving learners the knowledge and confidence to integrate data lake storage, master Delta Lake fundamentals, manage databases, and apply advanced techniques for data analysis, pipeline automation, and performance optimization.
Course code: PT1010
This Azure Databricks training teaches learners to process large-scale data and build machine learning models using Apache Spark, applied throughout to health insurance data such as claims, member, and provider records. Learners master Spark’s distributed architecture, query and transform data with Spark SQL and DataFrames, and apply Spark MLlib to build, engineer features for, and tune classification and regression models. The course closes with an introduction to streaming analytics, giving learners the practical skills to process both batch and real-time health insurance data at scale.
Course code: PT1011
This Azure Databricks training teaches learners to architect and manage unified data analytics platforms for health insurance organizations. Learners master the core components of the Data Lakehouse, including data ingestion, governance with Unity Catalog, and advanced visualization, applied throughout to health insurance data such as claims, member, and provider records. After taking this course, learners have the practical skills to build robust Medallion architectures, automate reliable claims and member data pipelines, and utilize AI-driven tools like the Azure Databricks Assistant to streamline operations and deliver actionable business insights for a health plan.
Course code: PT1018
Harnesses, Loops, and Graphs
Teams building an internal agent already know how to call a model. What they cannot explain twice is the run: whether it stops, whether a write happens once, and who can read what it did. This course is about the controls around that call. Over two days, participants wrap a tool-using task in a loop that stops, draw the same job as a map of steps including a person, and block a write until a check passes.
A harness is the rules around the model: what it may call, how many tries it gets, and when it must stop. A loop is try, check, stop. A graph is a map of steps, including a step where a person decides. A factory is that harness used on more than one job. The prompt asks. The harness decides whether the run may continue.
The course is for developers and technical leads reviewing a design before it touches production data. A guided track names the same steps in a builder. The build track writes them in code.
Course code: PT1019
Context Engineering for Business
Teams already answer from a policy, an account file, a project folder. What slows the work down is starting from a blank chat and hoping the model remembers the right documents. This course is about the material you put in front of the model before you ask: the files, the rules, and a short note on how your team works. Over two days, participants build a folder pack for a recurring job, get an answer that stays inside those files, and leave a way for the next person to run it.
Context, in this course, means that pack. People have heard the term RAG, a system that searches a library for you. Here the team chooses the documents and hands them over. If a folder is no longer enough, the last module is the request you would actually send to engineering.
The course is for people who work from documents and are being asked whether the team can do this inside the files it already has. No coding.
Course code: PT1020
Bridging Strategy and Value Creation
The work of this day sits with the person who decides what the team will run. Staff are already using chat tools on drafts, notes, and status. What is missing is a choice: which two workflows are worth funding, who owns them, who checks the work, and what would make you stop. The day is that choice, written as a plan a sponsor can fund or end.
An agent, in the language teams are already hearing, is a tool that can take steps toward a goal. The decision still belongs to the manager. This course is for that decision. People who need practice finishing a document or a deck are better served by a productivity class. This room is for the operating plan.
No coding, and no license, required. You bring one team workflow, with confidential details removed. The lab packet covers anyone who cannot bring one.
Course code: PT1021
Effectiveness for Technical Teams
Developers on this day already live in the editor, the repository, and the pull request. Copilot is already in that loop for many teams: a completion, a chat answer, or a proposed edit across files. What is uneven is the change that survives review. This course is about that change. Participants take a ticket through a plan, a bounded edit, a diff a teammate can read, and a team rule for when the assistant may touch files.
Agent mode, in Copilot, means the assistant proposes edits you accept or reject. Review still belongs to a person. The day uses the repository we provide, so no one brings production code. Where a Copilot license cannot be in the room, the same exercise runs in the lab editor and maps back to Copilot.
The course is for developers and tech leads who already use completions or chat and want assistant-authored work that passes review the first time.
Course code: PT1022
Quality Deliverables Made Efficient
Most of the work is already happening in a document, a slide deck, and a spreadsheet. What slows people down is the gap between a useful chat answer and a file a colleague can open. This day closes that gap. Participants take one set of meeting notes and carry it through the tools they already use: a document they would send, a short presentation built in the deck, and a simple dashboard they can reopen. The chat is used at each stage. The file is where the work is kept, checked, and finished.
The course is for working professionals who already write, present, and report, and who want that work to move faster without a technical detour. No coding. The labs run in a document editor, a presentation tool, and a spreadsheet.
