An interactive management system for the CFO organization's wholesale transformation, rendered as one double-ring loop: an AI-run inner machine (Foresight senses, Incubation pilots, Transformation scales) inside a thin human governance rim, run from an office of two with every other signature already on the firm's payroll. A business management system, not a technology program. Q1 2027 cycle; sample values flagged.

The delivery machine · the target operating model · 2027–2031 delivery program

A firm-scale mandate. An office of two.

This is a management system, not a technology program — the CFO organization's transformation run the way a portfolio is run. AI does the labor around the loop; two people steer it; and every signature that governs it already belongs to someone on the firm's payroll.

“As the delivery engine of the transformation, we convert AI capacity into the outcomes the firm has committed to — by running three disciplines that keep delivery FIT for purpose.”
Foresight — sense · Incubation — pilot · Transformation — scale.  Re-weighted here to run on agents.
Cycle
Q1 2027
Current priority
Charter № 2027-01
Next turn
in Q2 2027
Legend
sample
the outside — market · regulators · the Street · capability
Foresight
Sense
the market radar
⚙ agents draft
Incubation
Pilot
the portfolio & charters
⚙ agents sort
Transformation
Scale
the scorecard & close
⚙ agents reconcile
approve
sign
own
The mandate
Convert AI capacity into the outcomes the firm committed to — kept FIT.
Q1 2027 cycle
the loop turns weekly · closes quarterly
The inner ring is the machine — it runs the loop. The outer rim is the humans — thin by design: three gates, and the office adds only the first.
How to read this

One loop. The machine runs it; two people steer.

Scroll, and each discipline lights up on the ring and opens its live artifact here. Nothing on this page sits outside the loop: Foresight produces a market radar, Incubation a sorted portfolio and signed charters, Transformation a banked scorecard — and the reconciliation closes back to Foresight so the next turn is sharper than the last. Every artifact below is populated for the current Q1 2027 cycle; where only the inside view can supply a number, the cell is flagged sample.

F
Foresight · sense

The market radar — drafted by agents, blessed in one meeting.

Sensing agents read earnings, the Street, regulators and market structure, and draft dated signals — each one landing on a business flow, a date and a decision. The office does not research; it approves. ⚙ agents draft ✋ weekly review approves  → full Foresight instrument

7 approved signals · this cycle
Signal — latest readSo-what — the business decisionFlow · when
Row marks: what the outside demands · what moves the business · what funds the engine.
I
Incubation · pilot

The portfolio sorts itself — then one charter, signed.

A scoring agent ranks the firm's flows on three business tests. Move the weights — the tiers re-rank live; the office arbitrates disputes, not spreadsheets. Work-in-progress caps lighthouses at two: a small machine ships, it does not queue. ⚙ agents sort & assemble ✋ existing owners sign  → full Incubation instrument

Business flowCostReadyControlScoreTier
CHARTER № 2027-01 — Cross-border payment exceptions
Assembled by the charter agent · every action human-approved
Where the handle time sits sample
Intake & classification15%
Context — entity, sanctions, history35%
Repair drafting & rationale30%
Approval & release12%
Client communication8%

Agents target the red 65%. Baseline ~45 min, ~$28/case sample.

The bridge — dated, with a number to beat
Q1 27

Assisted triage live in one corridor, ~20% of volume; measure handle time, override rate, aging

Q2 27

Scale to ~60%; target cost per case −40% sample; first freed-capacity log

Q3 27

Redeploy-or-release decided, finance co-signed; banked as a share of the leverage review's payments row

Business owner: Services — cross-border ops Co-sign the bridge: Finance Clear the risk: Controls Delivery: the agents — no new hire
T
Transformation · scale

The machine closes the quarter itself.

Collector agents pull the metrics, assemble the scorecard and draft the reconciliation; finance co-signs what banks. Q2 2026 actuals are the honest baseline — advance the clock and watch the close fill itself. ⚙ agents assemble & reconcile ✋ finance co-signs

SIMULATED QUARTER — every filled value is sample°. The demo is the mechanics: agents collect, assemble, reconcile; finance co-signs. The first program close lands in Q1 2027.
AltitudeMetricLatest actuals · Q2 2026Status

The quarter-close reconciliation — latest-actuals shell (Q2 2026)

Deliberately a shell: a machine proposed mid-year cannot show a close it never ran. Advance the quarter to see the mechanics fill it.

✦
The human rim

Run from an office of two.

This is the answer to thin staffing — not a workaround for it. The machine is the team; the two humans steer; and everyone else who governs the loop already works at the firm. Low headcount is not the constraint the model fights — it is the model's whole premise.

2

The office — what only these two do

  • Principal — holds the mandate, chairs the weekly review, arbitrates what matters
  • Analyst — curates what the agents surface, preps the one pack the review runs on
  • That is the entire standing team. ~30 min sample a week is the steering cost.
+0

New hires — everyone else already signs here

  • The businesses own their flows and sign their charters — already their job
  • Finance co-signs what banks — already the returns-discipline role
  • Controls & Model Risk clear the risk tier — already the mandate they hold
The machine is the rest of the team.

That is how a firm-scale transformation mandate runs — at the highest autonomy the control environment allows — without standing up a program office. And the sunset test holds: when the unit-cost curve stays flat for four quarters without the office touching it, even the two step back.