DronaHQ Agents support multiple LLM providers and models through the built-in DronaHQ key. You can view the updated list of supported models and their credit usage here.

Usage is metered in credits based on the total tokens processed by the model. Each supported model has a defined credit cost per one thousand input tokens and per one thousand output tokens. Details and calculation examples are available here.

Agents use the built-in DronaHQ key by default. Bring your own LLM key is available on the Enterprise plan.

Agents can automate tasks across content generation, summarization, data transformation, CRUD operations, workflow automation, external tool integrations, knowledge base retrieval, and multi step orchestration. Any task that involves thinking, doing, or integrating can be modeled as an Agent.

Yes. You can mix providers and models across Agents. For example, you may use a light model for speed, a reasoning model for complex workflows, or an embedding model for semantic search.

Yes. DronaHQ provides scoped permissions, role based access control, version control, and secure connections to internal systems. These controls support compliance, traceability, and safe usage across sensitive environments.

No. DronaHQ enforces strict data use policies. Your Agent data is not used to train or fine tune any model. Contractual safeguards are in place for all supported providers.

Every Agent run is logged with full traceability, including inputs, outputs, tool calls, and timestamps. You can review these details in the Traces section. Trace data is retained for ninety days.

Yes. Each run includes a step by step breakdown of all tools invoked with their parameters and responses.

Yes. You can restrict topics, set strict instructions, limit tool access, and validate outputs. Guardrails will be available as part of the Enterprise plan.

Temperature adjusts creativity. Lower values produce more consistent responses. Higher values introduce more variation and creative output. You can tune this based on the use case.

Yes. Agents currently support image uploads for analysis. Support for more file types will be added in future releases.

Uploaded files are stored securely on DronaHQ’s Amazon S3 storage and cannot be accessed again after the chat is closed.

A Knowledge Base is a set of documents or information you attach to an Agent so it can reference accurate, organization specific content. This allows the Agent to use your internal material while answering queries.

The Agent can crawl content from the URL based on your crawl settings. You can set crawl depth and maximum URLs to control how many nested links are followed and indexed.

Yes. You can select an embedding model from the supported list to optimize retrieval for accuracy, speed, or cost.

No. DronaHQ Agents are currently available only on the cloud. Self-hosting options are not supported at the moment.

Short-term memory operates only within a single Agent run and is cleared after completion. Long-term memory persists across runs to support continuous or personalised experiences. You can enable or disable this based on your requirements.

Yes. Agents can use your existing connectors from the global connector library to work with your apps, databases, and APIs.

Yes. You can write JavaScript or Python and expose it as a custom tool for the Agent to use during execution.

Yes. Branding controls are supported. You can configure visual and identity elements to match your environment.