AI Agents Foundation & Practitioner
This course introduces delegates to the principles of building and managing AI agents that can perform tasks autonomously across business workflows.
Description
The AI Agents Foundation and Practitioner Course in United Kingdom helps professionals understand how autonomous AI agents can plan, reason, and perform tasks across digital workflows. It introduces practical approaches for designing intelligent agents that analyse information, interact with software, and support business processes.
- Build knowledge-powered AI agents using Retrieval-Augmented Generation (RAG)
- Design AI agents that can plan tasks, make decisions, and complete multi-step workflows
- Connect AI agents and web platforms with AI Agents Foundation & Practitioner Course
About This Course
Module 1: Architecture of Autonomy
- Agentic Frameworks
- What Makes an AI System “Agentic”
- Differences between Chatbots, Workflows, and Agents
- Event-Driven vs Goal-Driven Agents
- State Management in Autonomous Systems
- Memory Layers within Agent Architecture
- Linear Prompts vs. Autonomous Loops
- Single-Response Prompt Design
- Multi-Step Task Execution
- Feedback Loops and Iteration Cycles
- Error Correction within Loops
- When to use Linear vs Autonomous Designs
- The Plan → Act → Observe → Reflect Cycle
- Task Decomposition Techniques
- Planning Strategies for Complex Objectives
- Observation and Output Validation
- Reflection and Self-Correction Methods
- Iterative Refinement Logic
- Decision Trees in AI
- Designing Structured Decision Logic
- Branching Conditions in Workflows
- Deterministic vs Probabilistic Decisions
- Escalation Triggers
- Risk-aware Branching
- Giving Agents Structured Reasoning Steps
- Instruction Layering
- Constraining Reasoning Paths
- Preventing Reasoning Drift
- Output Formatting Enforcement
- Solving Multi-Part Problems Autonomously
- Breaking Complex Tasks into Subtasks
- Task Prioritisation Methods
- Dependency Handling
- Result Consolidation Strategies
- Practical Lab: The Logic Map
- Mapping an Existing Manual Business Process
- Identifying Decision Points Suitable for Automation
- Designing Conditional Logic Flows
- Defining Failure and Fallback States
- Designing the Agent's Autonomous Decision Boundaries
- Creating a Structured Agent Blueprint
Module 2: RAG & The Knowledge Layer
- Vector Databases & Retrieval-Augmented Generation (RAG)
- What Vector Embeddings Represent
- How Similarity Search Works
- When to use RAG vs Fine-Tuning
- Knowledge Grounding Strategies
- Accuracy Improvement Through Retrieval
- Giving AI a Long-Term Memory
- Short-Term vs Long-Term Memory Concepts
- Memory Persistence Strategies
- Session-Based Memory vs Database Retrieval
- Context Compression Techniques
- Semantic Search vs. Keyword Search
- Embedding-Based Search Logic
- Relevance Scoring
- Query Optimisation Techniques
- Precision vs Recall Trade-Offs
- Data Preparation for Agent Consumption
- Cleaning Unstructured Data
- Chunking Large Documents
- Removing Redundancy
- Standardising Document Formatting
- Structuring PDFs for Retrieval
- Extracting Clean Text
- Removing Noise
- Formatting for Chunk-Based Indexing
- Improving Retrievability
- Formatting Internal Wikis and Documentation
- Creating Structured Headings
- Designing Searchable Documentation
- Creating Q&A-Friendly Content
- Version Control and Updates
- Practical Lab: The Company Brain
- Ingesting Large-Scale Documentation
- Cleaning and Chunking Documents
- Creating Embeddings and Indexing Logic
- Building a Knowledge Retrieval System
- Testing Retrieval Accuracy
- Achieving Accurate Question-Answering Across Complex Source Material
Module 3: Tool-Calling & Actionability
- APIs for Non-Coders
- What APIs are in Simple Terms
- Request and Response Structure
- Authentication Basics
- JSON Structure Overview
- Rate Limits and Error Handling
- How Agents Communicate with External Software
- Function-Calling Logic
- Trigger-Based Actions
- Passing Structured Data
- Validating External Responses
- Connecting to Platforms such as Salesforce, HubSpot, and Shopify
- CRM Automation Scenarios
- Lead Management Workflows
- Updating Records Autonomously
- E-commerce Automation use Cases
- Multi-Agent Orchestration
- Role-based Agent Design
- Hierarchical vs Collaborative Agents
- Task Delegation Models
- Conflict Resolution Between Agents
- Designing Manager and Specialist Agent Roles
- Supervisor Agent Logic
- Task Routing Mechanisms
- Quality Assurance Layer
- Coordinating the Researcher and Writer Agents
- Research Task Definition
- Data Validation Checkpoints
- Content Generation Alignment
- Output Consolidation
- Practical Lab: The Automated Researcher
- Building a Three-Agent Team
- Designing Role Responsibilities
- Lead Identification and Company Research
- Structured Data Extraction
- Automated Personalised Email Drafting Within a CRM
- Testing Multi-Agent Coordination
Module 4: Deployment, Safety & Monitoring
- Guardrails & Ethics
- Defining Operational Boundaries
- Ethical Usage Constraints
- Output Moderation Layers
- Access Control Principles
- Preventing Prompt Injection Attacks
- Recognising Injection Attempts
- Input Validation Strategies
- Restricting External Instructions
- Sandboxing Untrusted Content
- Eliminating Hallucination and Unintended Actions
- Output Verification Workflows
- Confidence Scoring
- Cross-Check Mechanisms
- Safe Fallback Responses
- Human-in-the-Loop (HITL) Design
- When Human Approval is Required
- Escalation Thresholds
- Manual Override Mechanisms
- Accountability Logging
- Defining Critical Action Thresholds
- Financial Transaction Limits
- Data Modification Controls
- High-risk Decision Triggers
- Escalation Policies
- Building Permission and Approval Checkpoints Into Agent Workflows
- Role-based Access Design
- Workflow Approval States
- Audit trail Creation
- Compliance Reporting Readiness
- Practical Lab: Live Deployment
- Deploying an Agent Into a Sandbox Environment
- Configuring Guardrails and Permissions
- Stress Testing With Complex Customer Scenarios
- Simulating Edge Cases
- Monitoring, Logging, and Iterating on Agent Behaviour
- Refining Based on Observed Failures
Assessment
To achieve the AI Agents Foundation and Practitioner Certification, candidates will need to sit for an examination that measures their capability to design, configure, and implement AI-driven workflows and agents. The assessment focuses on applied knowledge of prompt engineering, automation logic, integrations, and safe deployment of AI solutions to solve operational and technical business challenges. The exam format is as follows:
- Question Type: Multiple Choice
- Total Questions: 60
- Total Marks: 60 Marks
- Pass Mark: 75%, or 45/60 Marks
- Duration: 75 Minutes
Who Should Attend?
This AI Agents Foundation and Practitioner Course equips delegates with the knowledge required to design and manage AI-driven automation systems. The training is particularly valuable for professionals who want to move beyond using AI tools and begin developing structured AI workflows, including:
- Product Managers
- Innovation Leaders
- Business Analysts
- Operations Managers
- Automation Specialists
- Digital Transformation Professionals
What's Included?
- World-Class Training Sessions from Experienced Instructors
- AI Agents Foundation and Practitioner Examination
- AI Agents Foundation and Practitioner Certificate
- Digital Delegate Pack
You’ll also get access to the MyTKA Training Portal, which will be your go to hub for all your training.
Hands-On Labs: Included as part of our online instructor-led delivery, these labs provide real-world exercises in a simulated environment guided by expert instructors to enhance your practical skills.
Prerequisites
There are no formal prerequisites for attending the AI Agents Foundation and Practitioner Course. However, basic familiarity with AI tools, digital workflows, or business process automation will support better understanding during the training.