Frequently Asked Questions

Explore answers to the most common questions so you can quickly find the information you need and move forward with confidence.

Request a Demo

Getting Started

  • What is Savvi?

    Savvi is an enterprise-grade AI platform built for financial services. Savvi enables banks, fintechs, and payments companies to build AI Apps and autonomous AI Agents using their own proprietary data, ensuring full ownership of their AI products at a fraction of the cost of building custom applications. 

  • Who is Savvi for?

    Savvi is purpose-built for banks, credit unions, fintech, and financial services teams who need transparent, auditable AI and ML use cases that meet strict regulatory standards.

  • How is Savvi different from other enterprise AI solutions?

    Savvi delivers enterprise-grade AI at a fraction of the cost and complexity of big ML platforms. There’s no requirement for a sophisticated AI team or a heavy tech stack—you can start with the team and data you already have.

    • Full AI ownership: You own every model you build with Savvi. Your data never leaves your secure environment and is never used to train anyone else’s system.
    • Built for finance: Transparency, auditability, autoscaling AI infrastructure, and centralized governance are purpose-built for fintech and financial services.
    • Flexible entry point: Whether you’re starting from zero or extending an existing data science AI Center of Excellence, Savvi meets you where you are and scales with you.
  • What real-world problems can Savvi solve?

    Savvi fintech customers have successfully solved gnarly problems like decreasing ACH debit returns, ACH risk scoring and decisioning, recommending optimal payment methods, and reducing transaction fraud. Banks have worked with Savvi to solve problems like identifying deposit churn risk, predicting lending risk, and deploying loan service agents. Learn more about how Savvi solves real-world problems for payment providers and banks

  • How do we get started?

    Savvi makes onboarding simple, whether you have an AI team or none at all. Most customers begin with a short pilot to prove value on a single use case, leveraging our existing AI App templates for fraud detection, payment optimization, credit offer optimization, loan performance forecasting, churn prediction, and more—then move to production in weeks, not months. 

    If you prefer, our professional services team works closely with you to identify AI use cases that align with your business goals. Then we will help you collect and integrate data, train the models, and optimize the models, resulting in production-ready AI Apps that can work everywhere from spreadsheets to your software products, websites, and even your chatbots. Contact Us today!

Features and Usage

  • Can we build our own AI Apps using Savvi?

    Yes. Use Savvi to build and deploy custom AI Apps and AI Agents with your proprietary data while retaining complete control and IP ownership.

  • How does Savvi support rapid experimentation?

    Data scientists and analysts can run multiple modeling experiments in parallel, with integrated monitoring and rollback for fast iteration.

  • How does Savvi fit into your existing stack?

    Savvi is designed to run as a single tenant solution in the cloud, ensuring regulatory and SOC 2 Type 2 compliant data, API, and model hosting and isolation. If desired, the Savvi AI Client Containers can be wrapped within your existing VPC or even deployed into your existing cloud-hosted VPC or on-prem solution.

  • How does Savvi ensure models improve or adapt over time?

    Savvi’s platform includes continuous-learning pipelines that automatically feed real-world outcomes back into each model. Performance metrics are tracked in real-time, and decision thresholds update automatically to handle new patterns—such as fraud tactics, payment-flow changes, and shifting customer behavior—without the need for a large in-house data science team. You get models that adapt as your business and the market evolve, with zero extra DevOps or manual retraining.

  • How fast is Savvi’s decisioning / response time & scale?

    Savvi supports auto-scaling in its client container architecture. It's designed for fast real-time decisioning (<40ms response times for real-time decisions in many cases).

  • What kind of data connections and integrations are supported?

    Savvi supports over 400+ prebuilt data connections to popular data systems. Also supports CSV/Excel uploads, REST APIs, and JS tag integrations. It has built-in flexibility, whether you're bringing clean structured data or less tidy/unstructured sources.

  • What governance/explainability features does Savvi include?

    Savvi provides transparency into any model-driven prediction, classification, or decision (why did the model make a particular decision?), audit trails of decisions, the ability to inspect inputs (features) used, and set business guardrails so machine learning outputs stay within acceptable risk bounds.

Security and Controls

  • How is customer data protected?

    Savvi AI is a single tenant solution. All data, models, and APIs are hosted within a secure and isolated client containers. The entire platform meets stringent security requirements, including SOC 2 Type 2, and is frequently pen tested to ensure we meet the highest security standards.

  • How does Savvi protect my data/model environment?

    Savvi uses isolated, single-tenant client containers. Your data, models, and APIs are segregated and secure. Data is encrypted at rest (using AWS KMS) and served via HTTPS with secure certificates (2048-bit RSA, SHA256).

  • Is Savvi SOC 2 certified, and how often are security audits or pen tests done?

    Yes, Savvi is SOC 2 Type II certified. Regular penetration testing is part of their security practice.

  • How does Savvi handle operational risk?

    Savvi AI has multi-tiered risk management built into the platform, from uptime and error monitoring with tools like NewRelic, to client-specific AI Model performance monitoring with the Savvi Health Check system, which provides continuous operational management of AI model performance to catch problems like data drift or model error monitoring.

  • How are API keys and access controlled?

    Each client will have their own set API Key, which are unique to those Savvi Client Containers. Access is controlled—keys are not resettable by outside parties, and must be used via secure channels. Session management uses modern cryptographic verification (JWT tokens).

  • Where is data processed/stored, and can I use a VPC or on-premise?

    Savvi supports hosting options with client containers; you can use your own VPC or an on-prem option. Data is stored encrypted and in isolated environments.

  • How does Savvi handle compliance with regulations in financial services?

    Savvi is built explicitly for regulated industries. Its architecture, data handling, auditability, and guardrails are designed to meet regulatory requirements (e.g., financial regulators, privacy, security). SOC 2 Type II certification, isolated client containers, and transparency in model behavior—all help meet compliance obligations. Savvi AI also provides detailed modeling reports that can be used to ensure and meet model governance requirements.