10-lesson course · about 14 hours
Build and supervise proactive agents that carry your project between reviews
For Consultants, project leads, managers and professionals who turn information into plans, decisions and client work
Everything for ProfessionalsSet up proactive agents around a real project for intake, evidence, actions, deliverables and reporting, with approved connections, human review and a clear way to pause or retire them.
What you’ll be able to do
By the end, you can…
How it works
A tutor that meets you where you are
- Lessons that adapt to your levelThe ten lessons follow one path from intake to reporting. The explanations, examples and practice adjust to where you are, so you spend your time on what you need, not on what you already know.
- An AI tutor you can ask anythingType or speak your questions while you work on your own project. Your tutor helps you write a job statement, design a test set or check a revision for dropped requirements, much as you would with a trainer sitting beside you.
- Quizzes and flashcards for each lessonShort quizzes and flashcards tied to each lesson help you check that ideas like traceability, update cycles and status criteria have stuck before you move on.
- Progress tracking and a certificateSee which lessons and practice outputs you have completed, pick up where you left off, and receive a certificate of completion when you finish the course.
Curriculum
The curriculum
The course runs in five modules, one for each agent and a final one that connects them over a project cycle. Each module has two lessons: one to build the agent and one to test it against the cases that usually go wrong.
Module 1Organize requests as they arrive2 lessons
1Define the request proactive intake agent
- Write a role name and narrow job statement for an intake agent, including what it must not do
- Choose between a single run and a continuing brief for your project and justify the choice
- Choose OpenAI Dots or Claude Duet after checking the tool's current features against your needs and your organization's rules
- Set the agent's permissions, approved connections, agreed report, human review points and pause and retire steps
2Test repeat and incomplete requests
- Build a test set of sample requests covering new, repeated, reworded, updated and incomplete cases
- Test whether the agent's processed record prevents repeated inputs from creating repeated tasks or messages
- Check that incomplete requests are flagged as open questions rather than completed with assumptions
- Revise the intake brief based on what the test shows and rerun it
Module 2Keep the project evidence current2 lessons
3Create an evidence-update proactive agent
- Define an approved source set and the sources the agent must exclude
- Set an update cycle and the conditions that trigger a report
- Design a source digest that records what changed, where, and which part of the assignment it affects
- Explain the difference between reporting real, relevant changes and reporting duplicates
4Review weak and conflicting findings
- Identify signs of weak evidence the agent should label, such as single, outdated or unclear sources
- Instruct the agent to present conflicting findings side by side without choosing between them
- Test the agent with seeded weak and conflicting examples and review its digest
- Record your review decisions and reasons in the project record
Module 3Keep actions and dependencies visible2 lessons
5Maintain the action register
- Build an action register structure that records the source of every commitment
- Set rules so only approved actions enter the register and missing owners or dates stay marked as not agreed
- Create an owner list the agent uses to match names, not to assign work
- Configure draft follow-ups that wait for your review before anything is sent
6Handle delays and blocked work
- Define status categories with clear criteria the agent applies consistently
- Explain how a delay in one action affects dependent actions in the register
- Choose which exceptions the agent escalates immediately and which go into the regular report
- Test the agent with delay and blocking scenarios and review its handling
Module 4Prepare the next useful deliverable2 lessons
7Build a deliverable-preparation proactive agent
- Build a requirement summary with identified, accepted requirements for the agent to draft against
- Define a repeatable drafting assignment: deliverable type, structure, audience and timing
- Set traceability rules so each section shows which requirement and evidence it draws on
- Explain why the agent flags gaps rather than filling them with plausible content
8Respond to reviewed changes
- Set the rule that only reviewed, recorded changes trigger a revision
- Build a change log and versioning approach for each revised deliverable
- Check a revision for hidden scope expansion or dropped requirements
- Test a late change across the deliverable, action register and open-question list
Module 5Test the proactive agents over a project cycle2 lessons
9Produce a recurring project review
- Build a shared project record with clear read and write permissions for each agent
- Produce a recurring progress summary and meeting pack from the shared record
- Run a compressed sequence of project updates and review the outputs of each cycle
- Choose the review content, cadence and recipients with the people who rely on the report
10Test interruption and resumption
- Pause each agent using the method your chosen tool currently supports, confirmed in its documentation
- Test resumption after an interruption for duplicate work, missed changes and repeated follow-ups
- Build a retirement and handover checklist for the agent set
- Review permissions, connections and reports for ongoing operation
Who it’s for
Made for you if…
Made for: Consultants, project leads, managers and professionals who turn information into plans, decisions and client work
Consultants handling client engagements
You turn unclear client requests into plans and deliverables, and you need agents that keep client-confidential records out of unapproved connections. You also need to keep evidence, judgment and approval separate.
Project leads coordinating several owners
You need a reliable picture of who committed to what and what is blocked, without agents inferring owners or deadlines and presenting them as agreed.
Managers who review work over time
You want a recurring review you can actually check, with clear permissions, agreed reports and a defined way to pause or retire agents when the work changes.
Professionals moving beyond one-off prompts
You already write clear briefs for AI tools and want to hand recurring preparation and follow-up to agents while keeping review responsibility with you.
Questions
Good to know
What do I need before starting?
You should have experience writing a clear, structured brief for an AI tool, or have completed the earlier curriculum days 1–2. You also need a project you recognize, with its requesters, owners, deadlines and open questions, and a basic awareness of how client-confidential and employee information should be handled.
Can I use my own project?
Yes. The course follows a sample service-improvement project, and you can substitute a comparable assignment from your role as long as you keep scope, commitments and review responsibility clear.
Do I need to write code?
No. The course does not cover custom code, APIs or software integrations, and it only goes into platform configuration as far as the lessons require.
Should I use OpenAI Dots or Claude Duet?
You choose one in the first lesson. Before you decide, you check the tool's current features against your needs and your organization's rules. Features change, so confirm them in the tool's documentation before relying on either, including how pausing works.
What will I produce?
Working outputs for your project: an action register, follow-up drafts, an owner list, a source digest, a requirement summary, a progress summary, an open-question list, an update draft and a meeting pack.
How long does it take?
Plan for about 14 hours across the ten lessons. It is self-paced, so you can spread the work around your project.
Are there live sessions or group classes?
No. An AI tutor teaches you one-on-one. Where a live class would rely on a trainer, you do the work yourself with the tutor's help.
Does this certify me in data protection or project management?
No. The course teaches you to keep sensitive records out of unapproved connections and to keep commitments visible. It does not provide legal, compliance or data-protection certification, or formal project management training such as PMP or Agile.