How we protect your data, handle evaluations safely, and work toward full compliance.
All evaluation data (scenarios, vendor profiles, agent traces, scores, and reports) is stored in PostgreSQL on Heroku. Data is encrypted at rest using AES-256. Each evaluation is isolated by customer ID.
We never store your credentials, API keys, or passwords. Sandbox execution environments use injected secrets that are ephemeral and never persisted. Your product is only evaluated through publicly accessible information.
Data retention is configurable per customer. By default, evaluation data is retained for 90 days. Enterprise customers can configure custom retention periods or request immediate deletion.
Agent Readiness Platform uses three LLM providers to run buying simulations. When we call these APIs, evaluation context (vendor names, URLs, publicly available product information) is sent to the respective provider.
We only send publicly available information (website content, pricing pages, documentation) to LLM providers. No private customer data is included in LLM prompts.
All automation is purely diagnostic. We simulate an AI agent's buying journey by reading publicly available web pages. We never create accounts, submit forms, make purchases, or perform any destructive actions on vendor websites.
Evaluations make a small number of HTTP requests to vendor websites (typically under 50 page loads per vendor per evaluation). This is comparable to a single human browsing the site and well below any rate-limiting thresholds.
We never attempt to log into vendor products, bypass authentication, or access protected areas. All evaluation is based on publicly accessible information: marketing pages, docs, pricing, and public API documentation.
The Agent Readiness Platform runs on industry-standard, trusted infrastructure:
Status: In progress. We are actively working toward SOC 2 Type II certification. Our infrastructure providers (Heroku/Salesforce) are SOC 2 certified. We expect to complete our own certification by Q4 2026.
We process minimal personal data. Evaluation data consists of publicly available product information. We will provide a Data Processing Agreement (DPA) upon request for Enterprise customers.
Our AI agent simulations follow responsible AI principles. Agents are instructed to behave as honest evaluators. All agent outputs are transparent and traceable through the run trace API.
Our team is happy to answer any questions about our security practices, complete security questionnaires, or discuss custom requirements for your organization.
Contact security@agentreadiness.com