Agentic AI for Professionals
A 12-week live program for professionals. You learn how AI agents work, build them with no-code tools, test them properly, and put the right controls around them. You finish with an agent built for a real process in your job.
- 12 weeks
- Two live sessions a week
- Live online
- No programming required
Overview
Agents that do real work, built by the people who know the work.
An AI agent doesn't just answer questions. It follows a goal across several steps, uses tools like your inbox, calendar and spreadsheets, and checks its own progress. This program teaches you to design those agents for processes you know, build them without code, test them properly and decide where a person must stay in control.
- Length
- 12 weeks
- Schedule
- Two live sessions a week: one to learn, one to build
- Format
- Live online, in small cohorts with an instructor
- Feedback
- A deliverable every week, reviewed by an instructor
- Capstone
- An agent for a real process in your own job
What you'll be able to do
- Explain how large language models and AI agents work, and where they fail
- Write structured, tested prompts for recurring tasks in your role
- Design an agent: its goal, tools, steps and where a person must approve
- Build agents that use email, calendars, spreadsheets and business apps without writing code
- Ground an agent in your organisation's documents so it answers with sources
- Test an agent with a written test set and measure its quality, cost and speed
- Apply privacy, security and human-in-the-loop controls to agents at work
- Make the business case for an agent and plan its rollout
Who it's for
Professionals, not programmers.
- 01Analysts, managers, project and operations people who want AI to take on real parts of their workload
- 02Professionals asked to lead or evaluate AI adoption in their team
- 03Consultants and business analysts who want to design agent-based solutions for clients
- 04Anyone comfortable with email, spreadsheets and business apps. No programming needed
Curriculum
Twelve weeks, from prompt to production.
Each week pairs a concept session with a workshop or lab, and ends with a deliverable an instructor reviews.
Week
Session 1
Session 2
Deliverable
Week 1
How large language models work
Session 1What LLMs are good and bad at: tokens, context windows, hallucination, and why the same prompt can give different answers
Session 2Workshop: map the tasks in your own job and sort them into automate, assist, or leave alone
DeliverableAI opportunity map for your role
Week 2
Prompting for real work
Session 1Structured prompts: role, context, examples, constraints and output format; prompt chaining
Session 2Workshop: build and test reusable prompts for three recurring tasks from your job
DeliverableTested prompt library (portfolio item 1)
Week 3
From chatbots to agents
Session 1What makes something an agent: a goal, tools, memory and a plan-act-check loop; agents vs. workflows vs. assistants
Session 2Workshop: take apart three real agent products and identify their goals, tools and failure points
DeliverableAgent teardown notes
Week 4
Designing an agent
Session 1Breaking a process into steps, deciding what the agent does vs. what a person approves, and writing clear instructions
Session 2Workshop: write a one-page agent specification for a process in your organisation
DeliverableAgent design spec (portfolio item 2)
Week 5
Tools and integrations
Session 1Connecting an agent to email, calendars, spreadsheets and business apps with no-code automation tools
Session 2Lab: build your first tool-using agent that reads an inbox and drafts replies for review
DeliverableWorking email-triage agent
Week 6
Grounding agents in your documents
Session 1Retrieval: how agents answer from your own policies, manuals and files instead of guessing; chunking and citations
Session 2Lab: build a document Q&A assistant over a set of company policies, with sources shown
DeliverablePolicy Q&A assistant (portfolio item 3)
Week 7
Multi-step and multi-agent workflows
Session 1Orchestration patterns: planner and worker agents, hand-offs, retries, and when a single agent is enough
Session 2Lab: build a multi-step workflow, such as qualifying an inbound lead and preparing a briefing
DeliverableMulti-step workflow with run log
Week 8
Testing and evaluating agents
Session 1Test sets, success criteria, spotting hallucinations and silent failures, and tracking cost and speed
Session 2Workshop: write a test set for your agent and run it; fix the three worst failures
DeliverableEvaluation report (portfolio item 4)
Week 9
Safety, security and governance
Session 1Data privacy, permissions, prompt injection, audit trails and human-in-the-loop controls
Session 2Workshop: risk-assess your agent and add guardrails and approval steps
DeliverableRisk assessment and guardrails plan
Week 10
Measuring value and driving adoption
Session 1Process metrics, time saved, quality and risk; building a business case; change management for AI tools
Session 2Workshop: write the business case and rollout plan for your agent
DeliverableBusiness case and rollout plan
Week 11
Capstone build
Session 1Capstone kickoff: choose a real process in your function and scope the agent with your instructor
Session 2Build lab: design, build and test your capstone agent, with an instructor checkpoint
DeliverableCapstone prototype and test results
Week 12
Capstone demo day
Session 1Final reviews and rehearsal: walk through design decisions, test results and guardrails
Session 2Demo day: present your agent, the evidence it works, and how it would be rolled out
DeliverableCapstone presentation and agent playbook (portfolio item 5)
Also included
- Built around your job
- Most exercises use processes from your own work, so the capstone is something you could actually use.
- No programming required
- Agents are built with no-code and low-code tools. Comfort with spreadsheets and business apps is enough.
- Testing and governance included
- Every agent you build is tested against a written test set and reviewed for privacy and security risks.
- Instructor feedback every week
- Each week ends with a deliverable that an instructor reviews.
What you leave with
Working agents, and the evidence that they work.
Everything you build is tested, risk-assessed and documented, so you can show it to your manager, a client or an interviewer and explain how it behaves.
Tested prompt library
Reusable prompts for the recurring tasks in your role
Agent design spec
Goal, steps, tools and approval points for a real process
Document Q&A assistant
An agent that answers from your policies and shows its sources
Evaluation report
A test set, results and the fixes you made
Risk assessment
Privacy, security and human-in-the-loop controls for your agent
Capstone agent and playbook
Built for your own function, demoed on the last day
Tools
What you'll work in.
Mainstream AI assistants and no-code automation, the tools most organisations already have or can adopt.
- ChatGPT, Claude and Microsoft Copilot
- No-code automation (n8n, Zapier or Make)
- Microsoft Copilot Studio
- Google Workspace and Microsoft 365
- Notion or Confluence for knowledge bases
- Spreadsheets for test sets and evaluation
Tuition
$899.00
Instalment plans are available at checkout.
- Do I need to code?
- No. Agents are built with no-code and low-code tools. If you can work with spreadsheets and business apps, you can do this program.
- How is this different from AI & Machine Learning?
- AI & Machine Learning teaches you to build and train models with Python. This program is for professionals who want to use, build and govern AI agents in their work, without programming.
- What will I build?
- Several working agents during the labs, and a capstone agent for a real process in your own job, with a test set, guardrails and a business case.
- Which AI tools do we use?
- Mainstream assistants such as ChatGPT, Claude and Copilot, plus no-code automation tools for connecting agents to email, calendars and business apps.
Not sure it's the right fit? Come to an info session.
We walk through the curriculum, the tools and what you'll build, with time for your questions.