You don't need new ideas; you need to rent out plumbing you already built. Below are five lines that each (a) make money near-term, (b) copy a proven model, (c) sit on an asset you already own, (d) fit 2026–28. Everything I cut is at the bottom — not as "no," but as "not yet — flips to yes when X." If you want to know whether I waste your time, read the cuts first.
Charge vetted partners for access to your 767k businesses (take-rate), instead of handing them outward discounts. Copy: Shopify App Store / AppExchange. No balance-sheet risk, minimal regulatory friction. A 12-partner pilot answers it in a quarter. The lowest-risk new dollar in the building.
Rent licences, treasury, FX and stablecoin to the fintechs/EMIs/PSPs you understand best; stablecoin as the settlement rail, not a consumer toy. The "banks won't trust a consumer brand" reflex is the exact AWS-era "retailers don't sell servers" reflex. The market is broken and uncontested — and you lived the pain. Build it through the US charter.
PRAGMA is an AWS-type asset: sell fraud/risk scoring and reusable embeddings to the same bank-for-banks segment; un-copyable (tens of billions of labelled events, 111 countries). AIR + the Revolut X MCP are a Kindle-type flip — the agent dissolves the menu. Real, owned, in production. But whether this is an AI-company premium or "a bank with a good AI model" is decided by the structural question below.
Sell (not SSO but) reusable, bank-grade identity on the 68m KYCs you've already completed; beachhead = visa-outsourcing (real, paying, broken). Revenue on identity you already verified. In a deepfake and AI-agents era, proof-of-verified-human appreciates. One paying beachhead, not SSO or "digital identity for everything."
Buy ageing IFA cash-flow (recurring fees + sticky HNW AUM), re-rate the cost base with your tech. This is M&A with people-integration — against your grain, not analytically forced — exactly the call Bezos had to override himself to make with Zappos. So do it as a cash-flow purchase (model like a bond, no premium), eyes open that it needs a founder's judgment-override, not a spreadsheet's permission.
Building PRAGMA clears the first bar — it's a real foundation model, not "we spent more on AI" (more IT spend ≠ becoming a technology company), done. The first step is in the org chart — when the new thing is named, and treated (as "additional services\department"?)
Is PRAGMA / AIR isolated and allowed to disrupt the bank — its own P&L, its own autonomy, AIR free to dissolve the app — or is it org-charted underneath the bank as a support function?
Christensen's 'innovator's dilemma' whole point (check the Apple's Jobs-Ive case-study): you cannot disrupt yourself from inside the efficient machine. Your efficiency is your superpower and the thing that smothers disruption, because disruption needs slack and redundancy that an efficiency culture (rightly) strips out. So the move isn't "be less efficient" — insane. It's: stay ruthless everywhere except one deliberately-redundant, isolated unit where the next S-curve is allowed to breed. That unit is where the $200bn-as-AI-company premium is won or lost — and where most competitors will quietly fail. It is also, not coincidentally, the single most useful seat an outsider could fill.
You filed de novo with the OCC (Mar 2026). Right structure, good move, respect. The trap that killed 2021–26 is still in the road, and it isn't a quality problem.
Seems like The Bank Holding Company Act drags the whole group into US consolidated supervision (UBO scrutiny, group-wide compliance harmonisation and reporting). Insulate structurally — either a (Nik's?) personally-held bank serving Revolut as a BaaS-client (the easier foreign-individual path, not the foreign-group gauntlet), or a staged stake in a BaaS-ready bank (Column / Lead / Vast?). Pitch the regulator their way — "a good bank we respect and barely change, plus a small team and tech to serve a few new business clients" (the client is you) — and split the entity into a boring banking half and a fast technology half. And you also unlock Pillar 02: correspondent banking only works from the US now. This isn't an idea; it's a map of a minefield I crossed.
None is "no." Each is "yes, when —":
Filter without mercy and only asset-backed moves survive — and notice the cuts aren't dead, they're triggered.
Two-three things you can't get from the people already in\around, who (rightly!) think only about Revolut and will tell you you're winning: (1) the scar 'outsider' tissue is real, (2) an adversary's-eye view from someone who actually built\uses your competitors' apps, not a consultant's deck about them, (3) just a quick test for any strategy meeting: whose rival apps are genuinely on your's phones — and used? (Usually, almost none. I could be wrong. As always.)
You have plenty of people who think only about you.