The three instruments — Foresight, Incubation, the leverage review — are quarterly reports, and a quarterly report should carry only what the quarter changes. Everything permanent was pulled out and put here. There is no quarter selector on this page — that absence is the point.
A record quarter, public medium-term targets, and a productivity agenda that now runs out of the CFO organization. These are the fixed points every cycle is argued against — they do not move when the quarter does.
The “can Citi fix itself” discount is closing. The conversation has moved on.
Every number is an operations statement: expenses growing materially slower than revenue, every year, through an investment cycle.
A portfolio with P&L accountability and “returns discipline” as the stated first principle — no longer a collection of technology projects.
The durability question. Q2's operating leverage was revenue-led — a +14% quarter does the heavy lifting, and a strong Markets tape flatters any ratio. The targets assume mid-single-digit revenue; the year Markets normalizes, the ratio is carried by the expense line or not at all. Separating the revenue-led from the productivity-led share of operating leverage — and growing the second — is the difference between a good quarter and a structural sub-55%. That is a business-transformation problem; technology is the instrument.
What each quarter's cycle does about it is the subject of the three instruments.
The wholesale franchise is three revenue engines sharing a single processing backbone. The backbone is where the addressable expense sits — and therefore where the radar points, every cycle.
Treasury & Trade Solutions (cross-border / international payments, USD clearing) + Securities Services. The highest-return, most process-heavy franchise — $416B of cross-border value moves through it.
Fixed Income + Equities (incl. Prime). High-volume, low-latency; exception- and reconciliation-heavy.
Investment Banking (M&A, ECM, DCM) + Corporate Banking. Document- and approval-heavy relationship work.
The shared spine — onboarding and KYC, reference & client data, trade & payment processing, reconciliations, settlement, controls, and regulatory reporting.
The discipline that governs every cycle: AI creates capacity; capacity becomes value only through explicit decisions. “Savings” is just one of the four currencies that value can be paid in — and it is not the one most of today's capacity is being paid in.
Released positions, avoided backfills, reduced vendor and BPO spend, location-mix shifts. The only currency that lands directly on the expense line — and the one severance has been buying. The transformation task: convert one-off severance into a structurally different shape.
Capacity moved into growth work: volume absorbed without hiring, expenses held flat while revenue compounds. Invisible as “savings,” decisive for the ratio — and the currency most of the ~100K weekly developer-hours are currently paid in.
Onboarding cycle time is time-to-first-revenue; payment certainty and exception latency are client experience — share of wallet in Services. Small percentages on a wholesale franchise running ~$15.3B a quarter (62% of firm revenue) outweigh large percentages on the cost base.
Cheaper, faster, audit-ready evidence and higher validation throughput protect the planned ~50% roll-off of transformation expense — the same roll-off that part-funds the investment envelope. “A large body of work passing internal audit validation” is a trajectory to defend, not just report.
The opportunities in Incubation are the supply side — where capacity comes from, and which of them is chartered changes every cycle. These five levers are the demand side — where capacity lands, on whose line, measured how — and they do not change. Sizing is gross, overlapping, one lever inside the committed program. Click any card.
Before the portfolio arithmetic means anything, the mechanic has to be concrete. Two flows, from copilot to multi-step agent — grounded in Citi data and bank-grade governance, with a human sign-off gate on any regulated action. Pick a scenario and run the agent.
A demonstration of the mechanic, not a benchmarked result — the timings are illustrative and the basis line names the external evidence each shape is drawn from.
A claim that something is “ready now” has to be backed by a result achieved somewhere, not a vendor promise. These are the documented quantum-leaps the whole system weighs — used as bounds on what is possible, never as forecasts for Citi. Including the failure, which is a design input too.
The modeled savings in Incubation sit deliberately below these bounds, and Klarna's over-automation of judgment work is precisely why human-in-the-loop tiers gate every regulated or judgment-dense step in this design.
Same rule across the system: a serious review will — and should — attack this. The challenges it expects, answered plainly — several conceded outright, because the concessions are the point. Click any card.
Nothing above moves when the quarter moves — which is exactly why it is not in the quarterly reports. Each instrument below carries only what its cycle changes: what the radar sees now, what is chartered now, what has banked now.