Skip to main content

API Keys & Model Provider Credentials (Company → Keys)

The Keys tab (under Company settings, alongside General, Security, and Models) is where you connect your own API credentials for third-party AI services and model providers. Adding your own keys lets you use custom models, enable real-time web search for agents, and manage your own usage/billing limits instead of relying on shared defaults.

Each provider card shows a edit icon to update credentials and a delete icon to remove them. Fields with -- are not yet configured.


Connect services like Tavily, Perplexity, or Parallel Web Systems to enable real-time internet search and retrieval capabilities for your agents. These services enhance agent responses with up-to-date information from the web.

ProviderFields
Tavily SettingsAPI Key
Perplexity SettingsAPI Key
Parallel Web SystemsAPI Key
Databar SettingsAPI Key

LLM Model Settings

Connect your own OpenAI, Anthropic, Gemini, or other model provider credentials to use custom models or manage your own usage limits.

ProviderFields
OpenAI SettingsAPI Key
Anthropic SettingsAPI Key
Gemini SettingsAPI Key
Groq SettingsAPI Key
OpenRouter SettingsAPI Key
Together AI SettingsAPI Key
Vertex AI SettingsService Account Credentials, Project ID, Location
Amazon Bedrock SettingsAWS Access Key, AWS Secret Key, AWS Region

Notes on multi-field providers

  • Vertex AI requires a Google Cloud Service Account Credentials file/JSON, plus your Project ID and Location (region) for the Vertex AI resources.
  • Amazon Bedrock requires standard AWS credentials: AWS Access Key, AWS Secret Key, and the AWS Region where Bedrock is provisioned.

Managing Keys

  • Edit (pencil icon) — opens the field(s) for that provider to add or update credentials
  • Delete (trash icon) — removes the stored credentials for that provider
  • Configured keys display as masked characters (e.g. ********) for security; unconfigured fields show --

Models Tab (Company → Models)

View the available LLM models and providers configured for your company. These models can be used for all new agents.

Each model card shows the model name, provider (e.g. OpenAI, Anthropic, Gemini, VertexAI, AWS Bedrock, OpenRouter, Together AI), a short description, and a status icon in the top-right corner:

IconMeaning
Green checkmarkModel is already enabled and ready to use in agents
🔑 Key icon + "Key required"Model needs an API key/credential before it can be enabled — go to the Keys tab and add the corresponding provider's credentials first
Plus iconModel is available and can be enabled directly — click it to turn the model on

Enabling a model that requires a key

  1. Go to Company → Keys
  2. Add the API key (or credentials) for that model's provider (e.g. VertexAI, AWS Bedrock, Together AI — see sections above)
  3. Return to Company → Models
  4. The model's icon will change from 🔑 Key required to ➕ Plus (enable) — click it to activate
  5. Click Save Changes to apply

Enabling a model that doesn't require a key

  • If the model shows a ➕ Plus icon directly (no "Key required" label), it's ready to enable with no extra setup — just click it and Save Changes Use the Search Models bar to quickly find a specific model by name.

Was this helpful?