Appearance
Reap product context
I grounded the prototype in Reap's actual product reality as of May 2026. This page is the research lens I used before deciding what to build.
Product lines
The way I read Reap: it is a stablecoin-powered financial infrastructure company, not a generic spend-management tool. The relevant product lines are:
| Product | What it is |
|---|---|
| Reap Direct | Flagship business account spanning fiat + stablecoin rails. Bundles cards, payouts, expense management. |
| Reap Card | Visa corporate credit card, collateralized in fiat or United States Dollar Coin. Multi-currency. |
| Reap Payments | Cross-border business-to-business payouts settled via stablecoins such as United States Dollar Coin. Volume engine: ~$3B/month business-to-business stablecoin payments in 2025 (30× growth from 2023). |
| Expense Management | Receipt capture through artificial-intelligence optical character recognition for vendor/amount/date, card-level policy rules such as merchant-category-code, geography, and time blocking, approval workflows, budgets, category reporting. |
| Embedded Finance | White-labeled stablecoin-native rails for fintechs, crypto exchanges, neobanks, PSPs. |
| Agentic Payments | A homepage marketing surface: "Enabling artificial-intelligence agents to transact on behalf of individuals and businesses." Treated as aspirational / directional, not shipped. |
Customer segment: small and medium-sized business → mid-market, Asia-Pacific-heavy, with Hong Kong headquarters. Web3-native businesses, import/export, payment service providers, crypto exchanges, neobanks. Not enterprise — no NetSuite story.
Corporate event: Payward, Kraken's parent company, announced a definitive agreement on May 7, 2026 to acquire Reap for up to $600M in cash and stock, pending customary approvals and expected to close in H2 2026. Strategic direction is "payments layer for stablecoins at scale" — finance-ops automation is a natural next layer.
Tech / data substrate
These are the events I assumed the prototype could plausibly hook into:
- Card transactions (Visa authorizations, with merchant category code, merchant, geography, and foreign-exchange data)
- Cross-border invoice payouts (fiat-in → United States Dollar Coin settlement → fiat-out)
- Foreign-exchange conversions between fiat and stablecoins
- Receipt artifacts (optical-character-recognition-extracted vendor, amount, date)
- Stablecoin on/off-ramp events (United States Dollar Coin settlement legs)
Accounting integrations: I could not find public evidence of Xero/QuickBooks Online/NetSuite-style accounting integrations on the expense page. My assumption is export-to-comma-separated-values at best. That gap is exactly why auto-tagging and ledger-ready payloads felt like credible prototype territory.
Artificial intelligence / agent positioning
- "Agentic Payments" homepage banner — directional, no shipped product.
- Expense page mentions artificial-intelligence receipt extraction through optical character recognition — the only concrete shipped artificial-intelligence feature.
- No public blog posts on "chief financial officer agent" framing.
My read: Reap is artificial-intelligence-curious, not artificial-intelligence-mature. A well-scoped agent prototype should feel genuinely novel for them, not me-too.
Competitive frame
Reap sits at an unusual intersection:
| vs | Where Reap differs |
|---|---|
| Brex / Ramp | Similar card + expense surface, but Reap's wedge is stablecoin settlement + Asia-Pacific/cross-border |
| Airwallex / Aspire | Closer fit for Asia-Pacific cross-border business-to-business, but Reap is stablecoin-native rather than correspondent-banking-native |
| Wise / Mercury | Reap targets businesses Wise/Mercury can't serve (web3, underbanked cross-border corridors) |
Wedge: stablecoin rails for businesses traditional banks will not touch, with a card + expense user experience on top.
How each skill maps to Reap's shipped surface
| Skill | Fit | Reasoning |
|---|---|---|
| Auto-tagging | Strongest | Reap already has card transaction streams and artificial-intelligence receipt optical character recognition but no public accounting-integration story. Multi-currency fiat and United States Dollar Coin classification plausibly extends their shipped surface and fills a real gap. |
| Policy enforcement | Medium | Reap already has rules-based card controls for merchant category code, geography, and time, plus approval workflows. A large-language-model policy agent risks duplicating shipped functionality unless framed as post-transaction policy review or natural-language policy authoring. |
| Accounts Payable agent | Good, higher stakes | The cross-border payouts product is the volume engine ($3B/mo). An accounts-payable agent that ingests invoices → routes via stablecoin rails → reconciles is on-strategy. |
Lead skill: I would lead with auto-tagging: an agent that classifies Reap card and stablecoin payout transactions across fiat/crypto, then emits clean ledger-ready entries for Xero/QuickBooks Online-style adapters. It uses shipped data, closes a visible product gap, and is demonstrable end to end.