Can I run Botlit agents on my own OpenAI or Anthropic account and keys?

Yes. Botlit[1] is built bring-your-own-model (BYOK) from the ground up and does not resell foundation model inference.

How it works

• 11 Provider Types: You can register your own accounts and API keys across 11 provider families, including OpenAI, Anthropic, Google AI, Azure OpenAI, AWS Bedrock, Mistral, Groq, Together, and custom OpenAI-compatible endpoints.
• Vaulted Credentials: Keys are stored securely in the platform integrations service at the workspace level. They resolve dynamically at runtime and are never hardcoded into agent definitions or exposed in plaintext.
• Connection Testing: Every provider integration includes a connection test to verify credentials, endpoint connectivity, and model access before agents bind to them.

Runtime execution

As detailed on the Bring your own model use-case page[2] and the AI overview page[3]:

• Standard responses execute using your vaulted provider keys, with token counts and costs logged to Botlit's run ledger rather than billed with platform markups.
• Botlit's execution engine runs OpenAI-wire protocols natively today. While OpenAI and Azure OpenAI support advanced workflows like agent delegation and tool loops out of the box, native SDK adapters for non-OpenAI wire formats (like Anthropic tool calling) are on the runtime roadmap.

To configure your keys, explore the Bring your own model guide[2] to set up your workspace provider connections.