7-lesson course · about 20 hours

Run a reporting Factory where a late correction reaches every page

For CFOs, controllers, FP&A teams, accountants and finance operations professionals

Everything for Finance teams

Coordinate three proactive AI agent teams through one shared record. Carry a late correction through analysis, narrative and slides, and give finance a pack to check, then release or return.

Try lesson 1 freeSee what you’ll learn

What you’ll be able to do

By the end, you can…

Organize three agent teams with named ownersAssign your existing agents to Data and Control, Planning and Analysis, and Management Reporting. Give each team a scope, inputs, outputs, a list of actions it must not take, and a named owner such as the controller, the FP&A lead or the CFO.
Write handoffs that refuse incomplete workDesign a shared record that holds the accepted dataset version, period, entity, currency, units, control status, assumption version and open exceptions. Write acceptance criteria so a receiving team rejects an incomplete handoff and records why.
Follow the approved record with Dots or Claude DuetPick OpenAI Dots or Claude Duet under your organisation's rules for nonpublic data. Set it to follow the record across handoffs and return reconciliations, variance explanations and unresolved differences, with measured variances kept apart from explanations that have no evidence yet.
Carry a late correction through the whole packTake a freight-accrual correction that arrives after the draft pack. Reconcile it into a new accepted version, mark dependent work as out of date, rerun only what it touched, return provisional text to its owner, hold the affected page and compare v1 with v2 in an impact log.
Show the Factory fails safelyTest changing inputs, blocked work and an interrupted run that restarts later. Confirm that the Factory stops, flags the problem and resumes from the record without double-counting or quietly reusing stale results.
Set the rules and checks finance signs againstWrite operating rules: permissions, stop conditions, retry limits and an exception-ownership map. Build release checks for reconciliations, assumptions, cross-pack tie-outs and named sign-off. Then run a full simulated cycle and record the controls needed before any further use.

How it works

A tutor that meets you where you are

  1. Lessons that adjust to your levelThe seven lessons follow the Factory from design to release. They adapt to what you already know, so a controller who has built reconciliation agents and an FP&A lead who has built variance agents each start from their own work.
  2. An AI tutor you can ask anythingType or speak your questions. Ask why a handoff should be rejected, test your acceptance criteria, or work through what the impact log should show after the correction lands. The tutor does the job a trainer would do in a live class.
  3. Quizzes and flashcards for each lessonShort quizzes and flashcards tied to each lesson check that you can apply handoff contracts, stale-marking rules and release checks. Passing on memory alone won't get you through.
  4. Progress tracking and a certificateTrack your progress across the three modules. You receive a certificate of completion when you finish.

Curriculum

The curriculum

The course runs in three modules that follow the syllabus's final three days: organize the Factory, carry a late change across the teams, then release a pack you can defend. Every lesson works on a simulated multi-entity, multi-currency close and never touches live ledgers or payment systems.

Module 1Organize the reporting Factory2 lessons

  1. 1Assign proactive AI agent team responsibilities
    • Assign each existing agent to Data and Control, Planning and Analysis or Management Reporting
    • Define each team's scope, inputs, outputs and list of actions it must not take
    • Explain which decisions stay with named finance owners
  2. 2Define the reporting handoffs
    • Write acceptance criteria for each team-to-team deliverable
    • Design a shared record that captures versions, status, open items, blocked work and approvals
    • Define how a receiving team rejects a deliverable and records the reason

Module 2Carry changes across the proactive AI agent teams2 lessons

  1. 3Connect the three proactive finance AI agent teams
    • Choose between OpenAI Dots and Claude Duet using your organisation's rules for nonpublic reporting data
    • Run the three teams from an approved reporting snapshot through the shared record
    • Test changing inputs, blocked work and a later restart, and record how the shared record shows each result
  2. 4Reconcile a late-period correction
    • Trace a late correction from intake through reconciliation, variance analysis, narrative, slides and dashboard
    • Test that cross-pack tie-outs catch any figure the correction did not reach
    • Explain how the shared record documents the correction and the new snapshot

Module 3Release a pack you can defend3 lessons

  1. 5Refine the operating rules
    • Write operating rules covering each team's scope, triggers, stop conditions and escalation
    • Specify data-handling rules for nonpublic reporting data and approved tools
    • Explain why the classroom system stays disconnected from live ledgers, bank accounts and payment systems
  2. 6Set the finance release checks
    • Build a release checklist covering reconciliations, open differences, evidence, assumptions and tie-outs
    • Define who approves assumptions and release, and how approval is recorded
    • Explain what blocks release and how blocked items are shown
  3. 7Run the full reporting simulation
    • Run a full simulated cycle from arriving files to a finance-approved pack
    • Test the cycle with at least one change, one blocked item and one restart
    • Document the controls and open risks a finance owner would need to review before further use

Who it’s for

Made for you if…

Made for: CFOs, controllers, FP&A teams, accountants and finance operations professionals

Controllers

Your reconciliation and intake agents work on their own, but nothing tells the downstream agents when an accepted figure changes. You'll design the Data and Control team and the handoff rules that stop an incomplete or superseded dataset from moving forward.

FP&A leads

A correction after the draft can leave the bridge, the slides and the commentary out of step. You'll set up Planning and Analysis to rerun only what changed and to send unevidenced explanations back to their owners instead of publishing them.

CFOs and finance owners who release the pack

You sign the pack, so you need proof that tables, charts, text and labels agree. You'll define the release checks, the approval record and what blocks release.

Finance operations staff and accountants

You've built and stress-tested single agents and want to see how they hold up as a coordinated system, with an owner for every agent and a person for every exception.

Questions

Good to know

What do I need before I start?

Days 1–7 of the Finance MasterCourse or the same skills. That means you can frame a variance question, recompute material totals and keep measured variances apart from hypotheses. You should also have built and failure-tested individual proactive agents (intake, reconciliation, variance review, driver update, scheduled pack) on OpenAI Dots or Claude Duet. You need working knowledge of the month-end close, confidence in Excel or a similar spreadsheet, and access to Dots or Claude Duet with permission to use simulated or approved sample data.

Does the course teach me to build the individual agents?

No. The Capstone starts from agents you've already built. Building intake, reconciliation, variance or driver agents from scratch is covered in days 3–7 of the MasterCourse. This course is about coordinating them as teams.

Should I use OpenAI Dots or Claude Duet?

Either one. The course teaches them as equals. In lesson 3 you choose one based on your organisation's rules for nonpublic reporting data. The course doesn't compare vendors or pricing.

Will I connect the Factory to our real ledger or bank feeds?

No. The course uses simulated multi-entity, multi-currency data. The classroom system stays separate from live ERPs, ledgers, bank accounts and payment systems, and lesson 5 explains why. Customer identifiers, account credentials and nonpublic reporting information stay out of unapproved cloud tools.

Can the agents post journals, approve budgets or make payments?

No. Preparation never authorizes payments, postings or trades. Finance approves assumptions and release, and a named finance owner releases the pack or returns it.

Does finishing the course certify our controls or give an audit opinion?

No. A simulation isn't an audit opinion, an assurance engagement or regulatory certification of controls. In the final lesson you record the controls and open risks a finance owner would need to review before any further use.

What will I have at the end?

A documented Factory design, handoff contracts, a shared record, operating rules, a release checklist and a full simulated reporting cycle. The capstone outputs include cross-pack tie-outs (XLSX), a review queue (PDF), a variance narrative (DOCX), financial slides (PPTX) and an exception-evidence folder.

How long does it take?

About 20 hours, at your own pace.

Finance AI Capstone: build, test and release from a reporting Factory

$397

One payment · lifetime access

  • 7 lessons in 3 modules: organize the Factory, carry changes across teams, release a pack you can defend
  • Hands-on work with OpenAI Dots or Claude Duet on simulated multi-entity, multi-currency close data
  • A late freight-accrual correction case with stale-marking, selective rerun and a v1/v2 impact log
  • An AI tutor you can ask anything, typed or out loud, with lessons that adapt to your level
  • Quizzes and flashcards for each lesson, progress tracking and a certificate of completion

Secure checkout by Stripe. Have a promotion code? Enter it at checkout.

Refunds: Full refund if you ask before you open your first lesson. Once you’ve started, purchases aren’t refundable, except where the law gives you that right.

Already bought it? Sign in