Spade Launches Predictions: A Transaction AI Foundation Model for Banks and Fintechs

Data and AI platform Spade has launched Spade Predictions, a transaction foundation model designed to help financial institutions analyze payment behavior. By moving beyond the 25-character merchant name limit of card networks, the tool allows banks of all sizes to predict customer needs and detect fraud with higher accuracy than traditional propensity models.

What was announced

Spade Predictions is a product suite built on a transaction foundation model that analyzes payment behavior alongside a verified merchant database. The tool is designed to provide regional banks and fintechs with the same modeling capabilities previously reserved for the world’s largest financial institutions. According to the company, more than half a dozen of the largest global banks have built similar internal models over the last 18 months, but these projects typically require dedicated research teams and years of governed history.

The suite operates through two core components: one that identifies businesses using Spade’s verified records rather than just payment messages, and another that interprets consumer spending patterns based on those real-world business identities. The platform includes three primary modules:

• Enrichment: Provides verified merchant, location, and brand data across all payment rails, including third-party-routed charges.
• Data Operations: Allows teams to query transaction data using natural language and receive answers in seconds, while automatically maintaining merchant identifiers (MIDs) as businesses rename or move.
• Predictions: Covers product propensity, customer segmentation, and fraud detection using merchant-level embeddings.

In testing for a US bank’s lending program, the foundation model outperformed traditional propensity models by 35%. When fine-tuned for fraud detection, the system replayed eight months of transactions and identified fraudulent merchants early enough to potentially block $3 million in spend, in one instance providing a warning five months before the issuer’s first recorded block. The system integrates via API or MCP server and relies on Spade’s Merchant Graph, which tracks 99.9% of US and Canadian businesses with sub-40 millisecond latency.

The companies involved

Spade is a financial technology firm specializing in transaction data and AI infrastructure for the banking sector. Based in New York, the company provides a platform designed to solve the "merchant name" problem—the historical limitation where card networks provide only minimal, often cryptic, text strings to describe a transaction. By maintaining a proprietary, verified record of nearly all businesses in the US and Canada, Spade allows financial institutions to see the actual brand and owner behind every payment rail.

The company’s market position is built on its Merchant Graph, which serves as the foundation for its data enrichment and operational tools. This infrastructure allows banks to move away from maintaining manual merchant lists and instead use automated identifiers that remain current as businesses move or multiply. Led by Co-Founder and CEO Oban MacTavish, Spade has positioned itself as a provider of transaction clarity, aiming to turn raw payment data into actionable insights for customer engagement and risk management without receiving personally identifiable information.

What FF News has reported before

FF News has followed Spade’s expansion into major banking ecosystems, recently reporting that Increase Integrates Spade to Deliver Real-Time Merchant Data Enrichment for Card Transactions. This followed another significant partnership where FIS Integrates Spade to Provide Real-Time Transaction Clarity for Debit Issuers, a move aimed at improving the experience for debit cardholders. Spade’s relationship with FIS dates back several years; the company was notably selected as part of the FIS Announces 2024 Cohort for Fintech Accelerator Program, Now in its Ninth Year, which provided an early platform for its merchant data technology.

What this means

This launch signals a shift in the transaction data market from simple "cleaning" to predictive intelligence. For years, the ability to build foundation models on payment behavior was a luxury reserved for Tier 1 banks with massive data science budgets. By productizing these models, Spade is effectively democratizing high-end customer analytics, putting pressure on regional banks to upgrade their legacy propensity models or risk falling behind in cross-selling efficiency. The significant improvement in fraud detection—identifying merchant risk months before traditional blocks—suggests that the industry is moving toward a proactive stance where the merchant’s identity is as important as the transaction amount.