Skip to course content
Northstar AI
Welcome, John Smith

Your learning module

Enterprise Agentic AI Specialization

12 modules · 135 lessons · browser-local progress
0% complete0 of 135 lessons

Course content

Modules and subtopics

Show by progress

Browse the complete 12-module curriculum

  1. Agentic AI vs Generative AI: Comparing Autonomy, Workflows, and Use CasesW1R1 · 20 min · Foundation readingContinue
  2. The agentic AI operating modelW1L1 · 15 min · LessonOpen
  3. Persistent instructions, skills, reverse prompting, and prompt contractsW1L2 · 20 min · LessonOpen
  4. Multi-agent orchestration, model routing, video-to-action, and parallel browser workW1L3 · 20 min · LessonOpen
  5. Multi-agent consensus, debate, and verificationW1L4 · 20 min · LessonOpen
  6. Context engineering, model routing, cost control, and scalingW1L5 · 25 min · LessonOpen
  1. Why MCP existsW2L1 · 5 min · LessonOpen
  2. The host-client-server architectureW2L2 · 5 min · LessonOpen
  3. Tools, resources, and promptsW2L3 · 5 min · LessonOpen
  4. Communication lifecycle, JSON-RPC, and transportsW2L4 · 5 min · LessonOpen
  5. Connecting to and using an existing MCP serverW2L5 · 5 min · LessonOpen
  6. Building a no-code MCP serverW2L6 · 5 min · LessonOpen
  7. Building a code-based MCP serverW2L7 · 5 min · LessonOpen
  8. Development environment, SDKs, and the build sequenceW2L8 · 5 min · LessonOpen
  9. Testing and debugging MCP implementationsW2L9 · 5 min · LessonOpen
  10. Authentication, authorization, and permissionsW2L10 · 5 min · LessonOpen
  11. Prompt injection, tool poisoning, and dynamic tool changesW2L11 · 5 min · LessonOpen
  12. Supply-chain security and secure operationsW2L12 · 5 min · LessonOpen
  13. Designing reliable MCP tools and workflowsW2L13 · 5 min · LessonOpen
  14. Deployment, scalability, and enterprise infrastructureW2L14 · 5 min · LessonOpen
  15. Agentic, multimodal, and real-time applicationsW2L15 · 5 min · LessonOpen
  16. Applied MCP case studiesW2L16 · 5 min · LessonOpen
  17. A hands-on learning pathW2L17 · 5 min · LessonOpen
  18. Community, packaging, and contributionW2L18 · 5 min · LessonOpen
  19. Ecosystem directions and advanced projectsW2L19 · 5 min · LessonOpen
  20. Mcp mastery assessmentW2L20 · 5 min · AssessmentOpen
  1. Rag foundations and the shift to agentic retrievalW3L1 · 10 min · LessonOpen
  2. End-to-end architecture for a multi-user RAG applicationW3L2 · 10 min · LessonOpen
  3. Building with AI agents - planning, context, validation, and version controlW3L3 · 10 min · LessonOpen
  4. From application shell to managed RAG - fast start and black-box limitsW3L4 · 10 min · LessonOpen
  5. Owning the retrieval pipeline - ingestion, embeddings, configuration, and securityW3L5 · 10 min · LessonOpen
  6. Maintaining knowledge quality - deduplication, metadata, parsing, and performanceW3L6 · 10 min · LessonOpen
  7. Improving retrieval quality - hybrid search, filtering, reranking, and tool loopsW3L7 · 10 min · LessonOpen
  8. Retrieving beyond documents - text-to-SQL, web search, and database-level controlW3L8 · 10 min · LessonOpen
  9. Whole-document reasoning with subagents and a transparent streaming interfaceW3L9 · 10 min · LessonOpen
  10. Testing, debugging, local operation, and production readinessW3L10 · 15 min · LessonOpen
  1. Why conventional RAG is not enough for stateful AI agentsW4L1 · 5 min · LessonOpen
  2. Represent memory with relational, vector, and graph layersW4L2 · 5 min · LessonOpen
  3. Turn raw data into searchable, structured memoryW4L3 · 5 min · LessonOpen
  4. Manage the complete memory lifecycleW4L4 · 5 min · LessonOpen
  5. Partition memory by domain, session, and agent roleW4L5 · 5 min · LessonOpen
  6. Combine vector retrieval with graph traversal for multi-hop answersW4L6 · 5 min · LessonOpen
  7. Use feedback to improve future retrieval and decision makingW4L7 · 5 min · LessonOpen
  8. Orchestrate multiple agents to resolve a cross-system incidentW4L8 · 5 min · LessonOpen
  9. Design for persistence, observability, latency, and deploymentW4L9 · 5 min · LessonOpen
  10. Understand the trade-offs and choose the right memory use caseW4L10 · 5 min · LessonOpen
  1. Start with a real workflow problemW5L1 · 5 min · LessonOpen
  2. Match the model, context, and interface to the workloadW5L2 · 5 min · LessonOpen
  3. Design goal loops instead of chasing perfect promptsW5L3 · 5 min · LessonOpen
  4. Decompose large work into specialized subagentsW5L4 · 5 min · LessonOpen
  5. Turn successful procedures into reusable skillsW5L5 · 5 min · LessonOpen
  6. Engineer context continuity with projects and shared memoryW5L6 · 5 min · LessonOpen
  7. Extend the agent with tools, but control the trust boundaryW5L7 · 5 min · LessonOpen
  8. Deploy recurring work where it can actually runW5L8 · 5 min · LessonOpen
  9. Make validation and recovery part of the build loopW5L9 · 5 min · LessonOpen
  10. Separate the model from the harness and compose an agent operating systemW5L10 · 5 min · LessonOpen
  1. The agent harness is the operating layerW6L1 · 5 min · LessonOpen
  2. Compress context and layer instructionsW6L2 · 5 min · LessonOpen
  3. Turn every run into persistent learningW6L3 · 5 min · LessonOpen
  4. Parallelize independent work and synthesize the resultsW6L4 · 5 min · LessonOpen
  5. Use stochastic consensus and debate to explore the solution spaceW6L5 · 5 min · LessonOpen
  6. Build lean specialist pipelines with fresh review contextsW6L6 · 5 min · LessonOpen
  7. Design autonomous improvement as a measured experiment loopW6L7 · 5 min · LessonOpen
  8. Choose the least complex automation layer that worksW6L8 · 5 min · LessonOpen
  9. Treat workspace organization as part of context engineeringW6L9 · 5 min · LessonOpen
  10. Contain security risk with simple controls and independent reviewW6L10 · 10 min · LessonOpen
  1. Version control, Git, and GitHubW7L1 · 5 min · LessonOpen
  2. The Git workflow and repository setupW7L2 · 5 min · LessonOpen
  3. Staging changes with controlW7L3 · 5 min · LessonOpen
  4. Commits, identity, history, and comparisonW7L4 · 5 min · LessonOpen
  5. Undoing, restoring, removing, and reversing changesW7L5 · 5 min · LessonOpen
  6. Branching for isolated developmentW7L6 · 5 min · LessonOpen
  7. Merging branches and resolving conflictsW7L7 · 5 min · LessonOpen
  8. Remote synchronization and pull requestsW7L8 · 5 min · LessonOpen
  9. Preserving unfinished work with Git stashW7L9 · 5 min · LessonOpen
  10. Rebasing for a linear historyW7L10 · 5 min · LessonOpen
  1. Building a local ComfyUI environment and model foundationW8L1 · 5 min · LessonOpen
  2. Understanding the node graph and the text-to-image pipelineW8L2 · 10 min · LessonOpen
  3. Controlling generation with resolution, seeds, steps, CFG, samplers, and denoiseW8L3 · 10 min · LessonOpen
  4. Organizing, reusing, and debugging workflowsW8L4 · 5 min · LessonOpen
  5. Extending ComfyUI with manager, models, and custom nodesW8L5 · 5 min · LessonOpen
  6. Upscaling images and controlling which outputs continueW8L6 · 5 min · LessonOpen
  7. Using image-to-image to balance preservation and changeW8L7 · 5 min · LessonOpen
  8. Producing detailed enlargements with tiled image-to-image upscalingW8L8 · 5 min · LessonOpen
  9. Controlling pose and composition with controlnetW8L9 · 5 min · LessonOpen
  10. Integrating external AI tools and consistent-face workflows with InstantIDW8L10 · 10 min · LessonOpen
  1. Manage the LLM context as a limited engineering resourceW9L1 · 5 min · LessonOpen
  2. Reach a shared design concept before requesting a planW9L2 · 5 min · LessonOpen
  3. Create a destination document without turning it into a specification-to-code contractW9L3 · 5 min · LessonOpen
  4. Turn the destination into vertical-slice issues and a dependency graphW9L4 · 5 min · LessonOpen
  5. Separate human planning from AFK implementationW9L5 · 5 min · LessonOpen
  6. Use test-driven development and feedback loops as the agent's control systemW9L6 · 5 min · LessonOpen
  7. Review in a fresh context and use manual qa to restore human judgmentW9L7 · 5 min · LessonOpen
  8. Shape the codebase around deep modules with simple interfacesW9L8 · 5 min · LessonOpen
  9. Control instructions with push and pull, and prevent documentation rotW9L9 · 5 min · LessonOpen
  10. Scale the workflow with planners, isolated implementers, reviewers, and mergersW9L10 · 10 min · LessonOpen
  1. Use an LLM council for consequential, ambiguous decisionsW10L1 · 5 min · LessonOpen
  2. Build the decision process around stochastic consensusW10L2 · 5 min · LessonOpen
  3. Create perspective diversity with specialized agent promptsW10L3 · 5 min · LessonOpen
  4. Install and register the LLM council skillW10L4 · 5 min · LessonOpen
  5. Operate the default single-model councilW10L5 · 5 min · LessonOpen
  6. Add model diversity with a cross-vendor councilW10L6 · 5 min · LessonOpen
  7. Configure cross-vendor access and protect the API keyW10L7 · 5 min · LessonOpen
  8. Run and inspect the cross-vendor workflowW10L8 · 5 min · LessonOpen
  9. Choose between default and cross-vendor modeW10L9 · 5 min · LessonOpen
  1. Move from tactical programming to strategic programmingW11L1 · 5 min · LessonOpen
  2. Optimize the harness, not only the modelW11L2 · 5 min · LessonOpen
  3. Design skills as reusable procedures without bloated contextW11L3 · 5 min · LessonOpen
  4. Use adversarial interviewing before implementationW11L4 · 5 min · LessonOpen
  5. Build stateful, mission-centered AI learning workflowsW11L5 · 5 min · LessonOpen
  6. Treat human knowledge, skill, and wisdom as the performance ceilingW11L6 · 5 min · LessonOpen
  7. Run scoped agents away from the keyboard inside sandboxesW11L7 · 5 min · LessonOpen
  8. Orchestrate agent work with queues, not endless loopsW11L8 · 5 min · LessonOpen
  9. Move human checkpoints rightward while preserving review and observabilityW11L9 · 5 min · LessonOpen
  10. Keep product vision and customer discovery human-ownedW11L10 · 5 min · LessonOpen
  1. Distinguish the model from the Hermes agentic harnessW12L1 · 5 min · LessonOpen
  2. Design Hermes as a self-improving AI employeeW12L2 · 5 min · LessonOpen
  3. Choose a runtime, cost model, and recovery pathW12L3 · 5 min · LessonOpen
  4. Grant computer access without surrendering controlW12L4 · 5 min · LessonOpen
  5. Build a multi-channel operating layerW12L5 · 5 min · LessonOpen
  6. Shape behavior with identity files and profilesW12L6 · 5 min · LessonOpen
  7. Engineer durable, time-aware memory with honchoW12L7 · 5 min · LessonOpen
  8. Add observability and deterministic project trackingW12L8 · 5 min · LessonOpen
  9. Scale from one agent to profiles, squads, and model routingW12L9 · 5 min · LessonOpen
  10. Connect external applications through MCPs, CLIs, and APIsW12L10 · 5 min · LessonOpen
  11. Turn repeated work into cron jobs, skills, and agentic workflowsW12L11 · 5 min · LessonOpen
  12. Use /goal for long-horizon, self-checking executionW12L12 · 5 min · LessonOpen
  13. Assemble a visual creation and deployment stackW12L13 · 5 min · LessonOpen
  14. Build a content production and distribution engineW12L14 · 5 min · LessonOpen
  15. Create personalized lead-generation and property marketing workflowsW12L15 · 5 min · LessonOpen
  16. Operate community management and customer support agentsW12L16 · 5 min · LessonOpen
  17. Build a paper-trading agent with separate execution and strategy layersW12L17 · 5 min · LessonOpen
  18. Build a private health accountability systemW12L18 · 5 min · LessonOpen
  19. Create a jarvis-style personal command centerW12L19 · 5 min · LessonOpen
  20. Productize Hermes systems as outcome-based servicesW12L20 · 10 min · LessonOpen