How does DryMerge communicate with my system?

  • DryMerge constructs workflows that call external APIs.
  • To integrate DryMerge with your system, you need to expose an API that DryMerge can choose to call directly.

Sending Data to Your System

  • For most standard webhooks, DryMerge will POST your end-user’s data to your systems’ APIs. The data will be in JSON format and will have the following structure:
{
 "for": "<service_name>",
 "user": "<your_end_user's_id>",
 "payload": {
     "compare": {
            "before": {
               // [your end-user's data as it was before the last check]
            },
            "after": {
               // [your end-user's data as it is now]
            },
            "diff": {
               // [the difference data before the last check and now]
            },
           "reverse_diff": {
               // [the difference data now and before the last check]
            }
        }
    }
}
  • We recommend exposing a single endpoint that DryMerge can post to. The endpoint should be able to handle the above JSON structure and extract the relevant data from it.
  • To tell DryMerge how to POST to this endpoint, simply tell it to in the Integration Config > Internal Prompt section of the dashboard. We recommend clearly stating the endpoint URL and telling it to POST to that endpoint by default.
  • This is ideal for customers who need the data for ingestion and indexing purposes due to its semi-structured nature.
  • If you need to POST to predefined endpoints, tell the LLM to wrangle the data into the format you need.

Receiving Data from Your System

  • Sometimes, customers want to construct workflows where they take internal data from your system and integrate it with their external tools like Google Sheets, Notion, etc.
  • To do this, you can decide to expose APIs that DryMerge can call directly to pull data.
  • To tell the chatbot how to call these APIs, simply tell it to in the Integration Config > Internal Prompt section of the dashboard. We recommend clearly stating the endpoints URLs and purposes as well as how to call them. OpenAPI schemas are very helpful here.
  • If your customers need to pull data from your system, but you don’t want to expose APIs, you can tell the LLM to pull data from your system and wrangle it into the format you need.

Running Commands on Your System

  • Customers may want to construct workflows in which they run commands on your system. For example, they may want to run one of your features on a schedule or through natural language (instead of calling the API directly/using a GUI).
  • To do this, you can decide to expose APIs that DryMerge can call directly to run commands.
  • To tell the chatbot how to call these APIs, simply tell it to in the Integration Config > Internal Prompt section of the dashboard. We recommend clearly stating the endpoints URLs and purposes as well as how to call them. OpenAPI schemas are very helpful here.