Anaplan CoModeler is built into the Anaplan modeling experience, and can be accessed when a model builder needs assistance. CoModeler reacts to natural language prompts. Just open your model and start a conversation. Work entirely through prompts, or add structured inputs to specify what you want. 

Part of a CoModeler conversation — adding a list

Note: Anaplan CoModeler is an entitled service and requires a separate subscription. Your Customer Success Business Partner can help you with this. You need to be a workspace administrator to use CoModeler. CoModeler is available in Polaris and Classic workspaces.

CoModeler ensures the user retains full authority over the model building process. The user specifies their intent, and CoModeler presents a clear set of proposed actions for approval. No changes are executed without explicit user confirmation. The platform provides real-time visibility into modifications, and maintains a comprehensive change log.

To open CoModeler and start a conversation, open a model and select on the upper left side of the page. The CoModeler panel holds the conversation. You have a Describe what to build field. This has to attach CSV file inputs to your conversation. Select to send your prompt to CoModeler.

CoModeler responds to your prompt in a conversational style. This conversation session remains active, and short-term context is retained.

You can download the text of the current conversation. Select .

When you start a new conversation, the context begins again. Select to start a new conversation.

In your interaction with CoModeler:

  • CoModeler tells you when it's thinking. Select Show process to see a commentary.
  • Use the keywords Approve or Approved to authorize CoModeler to make changes it has suggested. This could be an entire implementation plan, or, provided you specify exactly, parts of one. For example, Approve steps 5 and 6.
  • If you include Approve or Approved at the start of a prompt, CoModeler will just do it. You won't get an implementation plan for approval. For example, the prompt Approve create a module called Analytics creates the module immediately (unless CoModeler needs some clarification to continue).

See CoModeler tips for advice on how best to interact with CoModeler.

CoModeler operates within the permissions of the user who starts the conversation. CoModeler can't see, access, or modify parts of the model or data that the user themselves doesn't have permission to access.

CoModeler holds prompt-based conversations in English. It knows the current state of your model, and it remembers the last few prompts of the current conversation for context. At any point, you can start a new conversation. The power and responsibility to make judgments and decisions is yours. CoModeler only assists and fast-tracks your development.

If your request is unclear or incomplete, CoModeler asks clarifying questions before presenting you with an implementation plan. If you ask CoModeler to do something that it can't do, it'll tell you.

CoModeler only makes changes once you have approved an implementation plan. You can refine plans through conversation until you are happy to approve. Plans have numbered steps to make it easier to identify and discuss specifics.

Once you approve an implementation plan, if the result isn't quite what you want, you can change the model yourself, or prompt CoModeler further. You can revert the model to a previous version to undo changes. All changes made by CoModeler appear in the standard Anaplan model history, and are logged to the user directing the conversation.

Note: You can pre-approve prompts. Prefix your prompt with Approve or Approved to just do it. However, be alert to the risk in this.

CoModeler retains short-term context within a conversation. This enables follow-up prompts and clarifications. Once a conversation ends or a new one begins, the context resets. Conversations aren't saved between sessions.

Supplementing this content in Anapedia you have:

Anaplan has varied AI capabilities. Anapedia has content on Anaplan Agent Studio and Custom Analyst. This is how you bring AI agents to your application users.