Ramp Spend Management Solution

How we use AI

• 
  Our Security Team has also implemented data integrity, protection, and privacy protocols to ensure our use of customer data to keep data secure. We also have DPAs and contractual agreements with our AI subprocessors that restrict them from using or repurposing models or customer data provided by Ramp to improve their foundational models.
  
  Services provided to a customer, including AI-enabled services, are governed by our Platform Agreement, Privacy Policy, and Data Processing Addendum. If you have questions about how Ramp uses AI, please contact us at legal@ramp.com.
• 
  Finetuning of ranking models: Services utilizing this methodology solely output numerical similarity scores, which are not visible to the customer and are used to re-rank documents. These models do not output identifiable or other substantive information, which prevents the loss of any personal information. These are used to match a receipt to a transaction by ranking receipts by similarity to the transaction.
• 
  Finetuning of text-generation models: Ramp strives to keep privacy in mind by training its models with anonymized versions of customer data. Ramp uses a method of first removing identifiable information (e.g., name, email, phone numbers, etc.) and replacing the data with randomly generated strings. These could be used for services like memo suggestions for receipts, and auto-matching receipts to transactions.
  
  In-context learning (ICL) models: Ramp uses in-context learning with our models to perform inferences to service customers by referencing their existing data. The model does not retain customer data after use and customers cannot access other customers’ data or outputs. These are used to suggest merchant codes or transaction memos based on a customer’s historical patterns.
• Ramp is committed to building AI responsibly, meaning privacy, security, trust, and safety are central to what we do.
  
  AI is integrated into the Ramp platform to support an increasing number of Ramp services including Ramp Card, Ramp Bill Pay, and customer support. These tools were built to enhance the Ramp experience while maintaining the security and integrity of data. When training models, Ramp adopts methodologies to protect customer data and prevent the unauthorized exposure of customer information, including:

Corporate credit cards

• Set limits that are automatically enforced when employees spend—down to the category, merchant, and transaction level. Automatically collect receipts via text, email, and direct integrations with top merchants
• Powerful physical and virtual corporate credit cards with unlimited 1.5% cashback

Expense Management

• Ramp reviews every single expense the instant it happens and flags out-of-policy items for review.
• Reimburse employees for out-of-pocket expenses, made anywhere in the world, in less than 2 days
• Pre-approve expenses or issue benefits—with spend limits down to the daily level.

Bill payments

• When your bill is fully approved, the payment will go out automatically on the scheduled date.
• Create and customize approval workflows to implement a robust set of controls.
• Upload or email PDF invoices into Ramp and all the key fields will be extracted into a bill with 95% accuracy.

Accounting automation

• Build rules to automate coding, approvals, and reviews of transactions, while we collect receipts from your team automatically.
• Manage vendor data, bill payment details, and tracking categories and sync your data with just a click.

Reporting and insights

• Automatically find new ways for your business to save. Ramp surfaces actionable insights that highlight duplicate subscriptions, increases in spend, and unused partner rewards.
• Consolidate reporting with a macro view of expenses on cards, global payments & reimbursements on a single dashboard.

• - 1.5% unlimited cashback on corporate card

• - 8x faster to close your books each month

• - 90 seconds to pay a bill

• - 15,000+ customers (source: https://ramp.com/about-us)

• - $10B payments processed/year