Money20/20 USA 2026 – MissionOG Takeaways

 

Bottom line

With an ongoing technology debate as a backdrop, the trust debate has arrived and is ramping. Across stablecoins and AI agents, a question in every forum was who verifies the participant, who authorizes the action and who carries the loss when it goes wrong. This condition favors B2B infrastructure that sits beside banks: orchestration, identity and authorization, risk decisioning and dispute handling. The value accrues to the control layers, not the rails themselves.

 

1. Verified data is the source of truth

Every new capability on display, from agent payments to stablecoin settlement to AI underwriting, was gated by the same thing: data that can be proven, not just collected. Verified data is what unlocks automation, and owning it is the differentiation.

What we heard

• Merchants want certainty, not probability. Today’s fraud tools score risk, but merchants now need to know which agent is acting and who is behind it
• Data must travel with the payment.
• Consent is becoming per-request, not once per-session
• Point-in-time verification is failing. Sophisticated fraud with AI-built synthetic identities and deepfakes can pass a single check. What holds up is evidence that is costly to fake: behavior, device history, a real digital footprint
• Explainability is an audit requirement. A score with no reasoning behind it will get harder to defend to examiners, sponsor banks and customers

MissionOG lens

• Favor companies that produce or verify proprietary data as a byproduct of the workflow (permissioned bank data, payment-embedded data, consented network signals) over those that simply resell it
• Diligence question for every deal: is this company’s data verifiable, explainable and portable across rails and agents?

 

2. Agentic commerce: commercial payments first, trust layer is the prize

As expected, agents dominated agendas, and a general view was that they reach commercial payments before consumer checkout. Treasury and AP already run on approvals, invoices and vendor master records, which give an agent a mandate and an audit trail before money moves.

What we heard

• Mastercard projects up to 1 in 10 transactions could be agent-initiated by 2030
• Consumer checkout is not ready: in one panel test, shopping agents failed roughly a third of real purchases across ~1,000 merchant sites. Amex data shows 59% of consumers are comfortable with AI researching a purchase, but only 36% with an agent completing one
• The B2B proof points were real: Visa piloted trusted agents paying approved invoices with BILL and Billtrust using tokenized credentials, and announced supplier-validation and spend controls for commercial cards
• “Know Your Agent” was a key phrase of the week. Every agent payment must answer these questions: which agent, acting for whom, under what mandate and limit, and who is liable
• Like with many developing technologies, standards are fragmented: Visa Trusted Agent Protocol, Google AP2, Mastercard Verifiable Intent and Stripe shared payment tokens all compete. Difficult for a merchant to support them all alone

MissionOG lens

• Identity, delegated authority and fraud prevention are a critical gating layer, and an abstraction across competing agent protocols is a classic infrastructure wedge
• Disputes are the less obvious second-order effect (see Additional themes)
• Double check portfolio AP, commercial card and B2B payments companies on agent readiness now, before their enterprise buyers ask

 

3. Stablecoins in production, banks buying orchestration

Stablecoins moved from pilot to actual plumbing, and banks are paying attention to (spending) on orchestration rather than brand new rails. Both trends favor infrastructure built to partner with banks, not replace them.

What we heard

• A regulatory development occurred while the show was in progress: the OCC finalized the first GENIUS Act rule, setting how banks and non-banks become permitted stablecoin issuers (full reserves, redemption generally within two business days, effective in January)
• Payoneer surveyed 265 finance and treasury leaders: 69% already use stablecoins, 41% at meaningful scale. Yet when asked, only a handful of hands went up in a room of payments professionals who had personally paid with one
• Citi is seeing its strongest take-up in tokenized deposits, because clients never hold a wallet. Its view: tokenized deposits and stablecoins are “an and,” not a contest
• Mastercard launched Stablecoin Security, a risk score combining its fraud and identity data with blockchain analytics. BVNK (now part of Mastercard) wants real-world payments, not trading, to become the majority of volume
• Banks described the same gap: even with Zelle, ACH, RTP, FedNow, wires and push-to-card, there is no single system routing across them by cost, speed, approval rate and risk. Build vs. buy has become buy, then work to customize as needed
• Customers don’t ask for stablecoins per se; they want cross-border payments that move from point to point. The winning experience is invisible

MissionOG lens

• Multi-rail orchestration and compliance tooling (KYT, travel rule, wallet screening, fiat on/off-ramp) are the bank-side spend categories to bet on
• Rail-agnostic likely wins over stablecoin-native: products that treat stablecoins, tokenized deposits and RTP as interchangeable rails will capture bank budgets
• An open fight is yield. The OCC left rewards programs to case-by-case review and the ABA objected, so deposit-competition risk stays active for bank partners

 

Additional themes worth highlighting

• The agentic dispute wave nobody is pricing in

Chargeback rules now assume a human made the purchase. Agents break that: “my agent bought it,” “it misread me,” “I set a $150 limit and it spent $350,” “that wasn’t my agent.” None has a reason code today, and investigations get slower because nobody can easily reconstruct what the agent was told. Liability is shaping up as the next big network-rules fight, and stablecoin payments carry no dispute rights at all.

Why is it a concern: disputes, chargeback automation and evidence capture are a concrete, near-term spend category for issuers and acquirers, and a direct beneficiary of agent volume regardless of which protocol wins.

• Underwriting moves beyond the bureau

The credit pull still runs but no longer decides alone. Experian brought a cashflow data bureau, and Experian, TransUnion and Equifax debated a “new credit equation.” Lenders want more signal on thin-file and small business applicants but insist on explainable models they control.

Why is it a concern: cashflow and alternative-data underwriting, embedded SMB lending and explainable decisioning sit squarely in MissionOG’s B2B infrastructure thesis, and the bureaus moving in validates the category while raising the bar for standalone data vendors.