Skip to content

Building and Sharing Operators

For technical users, the platform offers rich programming tools. Developers can write Operator logic in code, using provided SDKs and APIs to interact with the system. For example, an API might allow an agent to take a screenshot, simulate keystrokes, or call cloud services. Modern frameworks already facilitate this: OpenAI’s new Agents SDK, for instance, helps developers coordinate multi-step tasks and tool use in agentic apps. Similarly, libraries like LangChain or custom SDKs can be integrated so coders have fine-grained control over agent behavior. The SDK also includes hooks for observability (logging agent actions) and safety (enforcing guardrails) so developers can test and debug their Operators.

Non-technical users are supported by a visual interface. A web dashboard lets users configure Operators without coding. For example, one could upload a video or screen-recording of a task and label the steps in the dashboard; the platform then trains an Agent to reproduce those steps. Users can set goals or rules (e.g. “alert me if this action fails”) via forms and wizards. This empowers domain experts to “teach” Operators by demonstration. Some RPA platforms already offer similar capabilities: for instance, recording desktop actions is a common way to build automations. Our platform takes this further by tying the recorded data to AI training, enabling Agents that improve and generalize.

The platform includes a curated marketplace to share and discover Operators. Creators can publish trained Operators (with metadata, usage instructions, and licensing) so others can deploy them directly on their systems. This fosters community collaboration: for example, a developer could create an Operator that automates invoice processing and share it; another company could deploy it in minutes. Big players are moving in this direction: Google Cloud’s AI Agent Marketplace, for instance, offers a catalog of pre-built agents for common business tasks. In our marketplace, token-based transactions (see Tokenomics below) could allow buying/selling of premium Operators. The system vets each Operator for security to ensure that only safe, compliant code is shared.