You have worked the five modules — the foundation, the monthly report, plan & forecast, the cash levers, and the close & the board. Each one ended with a graded artifact: the three foundation documents, a reconciliation and a leadership summary, a cash-flow model and a variance narrative, a collections plan and a vendor consolidation, a month-end-close runbook and a board pack. The capstone is where those five stop being five.
This is the integrative project that earns the credential. One company’s books, proven by closing and reporting one full month end to end — a complete finance operating system where every number traces back to the same clean export, in English and Arabic, from the categorization rule to the board figure. That is the thing “Certified Finance & Ops with Claude” attests, and it is a higher bar than five modules completed: a certified finance operator is not someone who watched the track, it is someone who can run a whole month’s close-and-report on a system that coheres and survives an auditor.
The capstone is not five assignments stapled together. It is one test: does the whole function hold as a single system — where the categorization rule you wrote in Module 1 is still load-bearing in the board figure in Module 5, and every number traces all the way back to its source row? A reviewer should not find a total in the board pack that drifted from the export it came from. Integration is the grade — coherence across the whole system, not five separate deliverables that each look fine alone.
The brief
Produce a complete finance operating system for one company, and prove it by closing and reporting one full month end to end. We recommend your own books — the capstone then doubles as a real close you ship and infrastructure you keep (see Bring your own books) — or the sample, Mizan’s own May close, the GCC bookkeeping SaaS whose finance team is threaded through the whole track, if you would rather learn on neutral ground.
You already know the Desktop motion: open the company’s folder in Claude Desktop, approve each read in the “Ask permissions” prompt, and assemble the system in the chat and the file pane. No terminal on this path. Claude drafts the totals and the narratives fast and shows its work on request; you trace every number to its source, and a person — never the preparer — signs anything that goes to the board, the FTA, or the bank.
Capstone deliverable — Close-in-a-Box (one company, one month, end to end)
1. The foundation (Foundation / M1)
- categorization-rules.md — every vendor → one bucket, every month
- data-checklist.md — profile-first, trace-to-source, show-the-work
- finance-controls.md — preparer ≠ approver, what's sensitive, bilingual
2. The monthly report, running (Report / M2)
- a reconciliation that ties the export to the bank
- a spend audit that explains the line that moved
- a leadership summary leadership actually reads — EN + AR
3. Plan & forecast (Plan / M3)
- a 6-month cash-flow model — every formula explained in one line
- a budget-vs-actuals variance narrative, tied to the spend audit
4. The cash levers (Cash / M4)
- an AR aging + a collections plan — who owes, how overdue, the chase
- a vendor-spend consolidation + a renewals calendar
5. The close & the board (Board / M5)
- a month-end-close runbook — the repeatable monthly checklist
- a board pack, every number traced to source — EN + AR narrative
6. A short reflection
- where Claude genuinely helped, and where your judgment was needed
- the ONE number you re-checked, killed, or rewrote because a person had to
Every piece must inherit #1. That inheritance is the whole assignment.The reflection is not a formality. Naming where you overrode the draft — the total you re-added by hand before you would trust it, the variance you would not sign until you had traced it to the row, the Arabic board line you re-authored because the translation read foreign — is part of the proof that you ran Claude well rather than letting it run you.
How it’s assessed — the master rubric
Five criteria, each scored meets / nearly / not yet. One “nearly” anywhere is a revise, not a pass — because a finance function that coheres in four places and drifts in the fifth still has one number nobody can trace, and an untraceable number is exactly what fails later: in the audit, in the board meeting, in the FTA filing. The bar throughout is a single question: would a CFO sign this number and put it in front of the board, the FTA, or the bank?
Master rubric — Close-in-a-Box
Each criterion: meets / nearly / not yet. The bar: would a CFO sign this
number and put it in front of the board, the FTA, or the bank?
1. Verified & traceable — every number traces to a source
meets Every figure in the pack names its source file and the rows
summed; spot-checks re-add by hand and match; nothing rests on
an unverified total.
nearly Mostly traceable, but one or two figures can't name their source
rows, or a spot-check doesn't reconcile.
not yet Numbers asserted without a trail; a total can't be re-derived.
2. The math is shown — a reviewer can re-run any figure
meets Every total and formula is explained in plain steps; the 6-month
model's logic is legible; a stranger could reproduce any number.
nearly The method is there but blurs in places, or one figure is a
black box.
not yet Numbers without method — re-running them is guesswork.
3. Used Claude well — drafted fast, a person decided
meets Claude did the heavy lifting, but every total and formula was
checked, the controls gate held (preparer ≠ approver; sensitive
data stayed in the workspace), and the reflection shows real
judgment overriding the draft where it mattered.
nearly Mostly sound, but a number slipped through unchecked, or the
judgment calls are asserted rather than shown.
not yet Reads as raw model output — confident, unverified, unsigned.
4. Arabic quality — an authored peer, not a translation
meets The leadership and board narrative read native: authored
register, Western numerals, AED, RTL tables, and they cleared
the same trace-to-source and sign-off gate as the English.
nearly Competent but translation-flavored, or the conventions slip.
not yet English run through translation; reads foreign; skipped the gate.
5. Completeness & coherence — one system, not six folders
meets All six parts present and connected: the rules feed the report,
the report feeds the plan, the cash levers feed the board, and
one number traces unbroken from the export to the board figure.
nearly A part is missing, or present but disconnected from the rest.
not yet A pile of deliverables, not a finance function.Notice that three of the five criteria — traceability, used-Claude-well, and completeness — grade the seams between the pieces, not the quality of any one of them alone. That is intentional. The modules already graded each artifact on its own; the capstone grades whether the rule, the report, the plan, the cash levers, and the board pack hold together as one finance function.
Bring your own books
Use your own books. We recommend it for one blunt reason: the capstone stops being coursework and becomes two things at once — a month you actually close and report, and infrastructure you keep. The system you assemble — the categorization rulebook, the data checklist, the controls doc, the reconciliation and spend-audit habits, the cash model, the collections and vendor playbooks, the close runbook and the board pack, in both languages — is real infrastructure your finance team keeps and runs, and the month you run it on is a close you actually ship. That is day-one ROI: the assessment, the portfolio piece, and the actual work are the same artifact.
The sample — Mizan’s own May close — is here for when neutral ground is the point: a new hire learning the system, or a controller who wants to practice the integration once before doing it for real. Mizan’s May is worked through every module — the 1,248-row export profiled in M1, reconciled and audited in M2, modeled in M3, chased and consolidated in M4, closed and presented in M5 — so you can study the moves on books with no quarter-end pressure attached, then turn around and do it for your own. Either path clears the same rubric.
What you earn
Clear the rubric and you earn “Certified Finance & Ops with Claude” — a verifiable page at /cert/[id] you can share and add to LinkedIn. It is the credential of the team track: proof that you can run a whole month’s close-and-report on a coherent system, not just describe the moves.
Here is the honest state of it today, because the credential is only worth what it honestly claims:
- A person grades it, now. In a cohort or concierge run, a reviewer scores your finance system against the master rubric above and tells you what is weak. The rubric and the senior-controller bar are the IP, and they are hand-gradeable today — that is the deliberate first step, not a placeholder.
- AI-assisted grading on the same rubric is next, not a claim we make yet. When it ships, it grades against this exact rubric — the bar does not move, the reviewer does.
- It is not an accreditation-body badge. Its weight is the work behind it: a complete, coherent finance function a reviewer judged against a senior-controller-and-CFO bar. That is a real thing to point at — and a more useful one than a quiz score.
No automated verification or paywall is wired behind this today that the line above does not describe. The certificate means what the work means.
What good looks like
The detailed examples live in the five modules. What the capstone adds is the through-line — the same traced numbers visibly threading every layer of one month. Here is Mizan’s May Close-in-a-Box at a glance, one or two lines per stage, so you can see the inheritance the rubric is really grading:
Mizan — Close-in-a-Box at a glance (May)
Through-line AWS → Hosting & infra, always — never "software." One rule,
applied every month, is what makes the trend lines real and
every board figure trace back. M1's rulebook is load-bearing
all the way to M5.
Foundation categorization-rules.md, data-checklist.md, finance-controls.md.
The checklist catches the dirt before any total: 1,248 rows,
3 blank excluded, 2 duplicate charge IDs investigated, 14 MM/DD
dates normalized. The clean export is the one source.
The report Reconciliation ties the 1,248 rows to the bank. Spend audit:
Software & tools jumped to AED 26,000 — and because AWS stayed
in Hosting (48,000), the jump is a real cost change (a new tool
+ a price hike), not a miscategorized invoice. Summary EN + AR.
Plan A 6-month model from the AED 2,400,000 opening balance, every
formula explained in one line. Budget-vs-actuals narrates the
Software variance against the spend-audit finding — same number,
same story, no drift.
The cash AR aging: Buraq Accounting AED 38,500 at 62 days is the chase;
the collections plan owns it. Vendor consolidation kills one of
the two overlapping project tools; renewals calendar built.
The board Board pack: ARR AED 6.1M, net +AED 40,000 for the month, runway
from the cash model — every figure traced to source, founder +
controller signed. Arabic board narrative authored — same gate.Read it top to bottom and the test is obvious: the rule from the foundation — AWS is hosting, never software — is still load-bearing in the board figure, because it is what makes the Software-line jump a real cost change you can explain rather than a categorization artifact you can’t. The net +AED 40,000 traces unbroken to the 1,248-row export it came from. Nothing drifted. That is what “meets” on traceability and coherence looks like — not six strong deliverables, but one set of numbers that survives all the way from the export to the board.
Now ship it
You have the brief, the rubric, and the bar. The capstone is not harder than the modules — it is the modules, run together for one month, with the seams made to hold. Pick the company, open the folder, and assemble the system Claude drafted and you traced and signed.
Two things to keep beside you as you build: the Foundation toolkit — the finance CLAUDE.md and the templates every later piece is built on, so the whole system inherits one source of truth — and, if you are rolling this across a team rather than certifying yourself, the operating guide, the data, controls, and sign-off layer that belongs underneath a team putting real money through Claude.
Build the system, clear the rubric, and the certificate is yours — because the work is.