runQC is a multi-agent adversarial QC platform for AI agents. It is in closed beta; invited teams run it on their own API key.
What an invited team receives during the closed beta
Access to the closed beta is by invitation.
BYOK means you provide your own OpenAI API key. runQC uses it to power the QC agents that test your AI agent.
It keeps the beta free for you and for us. You control your LLM spend directly. Typical test runs cost $0.01 to $1.75 depending on mode.
Your provider key is stored in AWS Secrets Manager, encrypted at rest and in transit. It is never logged or exposed. See Section 5 of our Privacy Policy for details.
Everything you need to know about the runQC beta
BYOK stands for "Bring Your Own Key." You provide your own OpenAI API key, and runQC uses it to power the QC agents that test your AI agent. Your LLM costs are billed directly by your provider. Typical runs cost $0.01 to $1.75 depending on the test mode.
Credits are how runQC measures platform usage. Suite test runs cost 1 credit. Intelligent test runs cost 2 credits. Invited beta teams receive 300 credits per month, which covers 150 to 300 test runs depending on the mix.
We'll introduce paid plans with additional features and higher limits. Your beta data will be preserved. Beta users who provided valuable feedback may receive extended free access or discounted plans.
Any AI agent accessible via a REST API endpoint. This includes chatbots, virtual assistants, customer service agents, specialized domain agents, and any LLM-powered system that accepts text input and returns text output.
Yes. Your provider API key is stored in AWS Secrets Manager, encrypted at rest and in transit. It is never logged, never exposed in the UI after entry, and never shared with third parties. See Section 5 of our Privacy Policy for full details.
An invited beta team's entitlement is: 300 credits/month, 2 concurrent test runs, 3 team seats, and data retention per our Privacy Policy. Credits reset monthly. There's a $5/day cost cap as a safety net for your API key usage.