Skip to content

Writer AI Agent

Writer AI Agent is a third‑party agent from Writer.com that uses a defined input schema to handle writing tasks. It specializes in generating, refining, and structuring content based on user input, leveraging integrated tools and customizable guidelines to support content planning, drafting, and iterative editing.

Use the AI Refinery SDK and the WriterAIAgent class to integrate your Writer AI Agent.

Creating Writer AI Agent

To set up an AI Refinery project with a WriterAIAgent, follow these steps:

  1. Obtain your API key and applicaion ID from Writer AI Studio:

    • Log in to the platform at https://dev.writer.com/.
    • Click Build an agent, then choose API to create a new agent.
    • Copy the Application ID once the agent is created, which is a UUID string.
    • Navigate to API Keys, find the Production key, and click Reveal key to display <your-writer-api-key>.
    • Store this value as you environment variable.
      export WRITER_AUTH_TOKEN=<your-writer-api-key>
      
  2. Configure api_key_env_var and application_id in your YAML configuration for AI Refinery

Onboarding Writer AI Agent

To use the Writer AI agents through our AI Refinery SDK, users need the following parameters:

Variable Description
api_key_env_var Name of the environment variable storing your Writer.com API key.
application_id Unique identifier for your Writer AI Agent provided by Writer.com.
wait_time (Optional) Maximum time in seconds to wait for a response.
contexts Allows for the provision of additional information during communication with the Writer AI Agent.

Workflow Overview

The workflow of the WriterAIAgent class consists of six components:

  1. Initialization
    An agent is created in Writer AI Studio and registered in AI Refinery with the api_key_env_var and application_id (see Usage and Quickstart).
  2. Requesting Input Schema
    AI Refinery requests the input schema from the Writer AI Agent.
  3. Preparing Input Schema
    The Writer AI Agent returns its input schema, and AI Refinery structures the request according to that schema.
  4. Sending a Structured Prompt
    AI Refinery sends a schema‑compliant writing prompt to the Writer AI Agent.
  5. Processing
    The Writer AI Agent uses its language model, built‑in utilities (e.g., grammar checking, summarization), and an optional knowledge base to generate content.
  6. Receiving the Response
    The Writer AI Agent returns a schema‑based output to AI Refinery for further integration.

Usage and Quickstart

To quickly set up an AI Refinery project with a WriterAIAgent, the user should first obtain their credentials from the Writer. Once the credentials are available, use the YAML configuration template below to integrate it into the AI Refinery project. Specifically, ensure the following configurations are included:

  • Add a utility agent with agent_class: SalesforceAgent under utility_agents.
  • Ensure the agent_name you chose for your SalesforceAgent is listed in the agent_list under orchestrator.

Template YAML Configuration of WriterAIAgent

See the YAML template below for the WriterAIAgent configuration.

orchestrator:
   agent_list:
      - agent_name: <your-agent-name> # Required, the name you choose for your WriterAIAgent  

utility_agents:
   - agent_class: WriterAIAgent  # Required and be "WriterAIAgent" for communication with user-established Writer AI Agent
      agent_name: <your-agent-name> # Required, the name you choose for your WriterAIAgent  
      agent_description: <brief description of functionality>  # Optional yet essential description of the utility agent outlining its functions.
      config:
      api_key_env_var: "WRITER_AUTH_TOKEN" # Required for authentication, the name of the environment variable
      application_id: <real-writer-application-id> # Required to access the unique application, actual ID from Writer.
      wait_time: 300 # Optional timeout in seconds
      contexts:   # Optional additional agent contexts 
         - "date"
         - "chat_history"

With this configuration, the WriterAIAgent will receive prompts, generate structured outputs, and return them to AI Refinery for further processing or workflow integration.