Sub-Workflow: Prevent Duplicate API Calls & Manage Data Flow

It seems your topic is missing some crucial details. Could you please provide the following information, if it applies?

• callin.io version:
• Database (default: SQLite):
• callin.io EXECUTIONS_PROCESS setting (default: own, main):
• Running callin.io via (Docker, npm, callin.io cloud, desktop app):
• Operating system:

I'm aiming for something similar to the below, although I recognize that the wait node cannot be used in this manner. This is merely to illustrate the concept. Essentially, the sub-workflow should begin with an initial state where data is loaded upon workflow activation. Subsequently, the data processing occurs as is standard practice, driven by the Execute Workflow Trigger. How can I achieve this or a comparable setup using the workflow and sub-workflow structure?

Great point!

For most use cases, I wouldn't be concerned about the API limits. Are you encountering quota issues with your current workflow?

Regardless,

I believe the simplest solution is to incorporate chat memory into the agent. This allows tool results to be saved as part of the ongoing conversation. If the agent determines it has sufficient information from a previous message, it might opt not to use the tool again, thereby reducing API requests.

An alternative approach would be to build your own caching system, perhaps utilizing a fast and lightweight solution like Redis.

It's worth noting that you can set an expiration time for cached data, giving you control over how stale it can become. In the example below, I've set the expiration to 5 minutes.

Thank you Jim, I believe I'll find both methods you provided useful. My current objective, however, is to develop a concise code example that exclusively utilizes callin.io nodes. I consider it good practice to bypass any superfluous API calls. If I were coding, I'd simply store it as a variable or read/write from disk. However, it appears that each execution in callin.io might discard the variable option? This leads to substituting your Redis approach with a callin.io node. But I'm unsure which node would be appropriate. Do you have any suggestions? Perhaps the in-vector memory store? Though this might not be as "general" as I'd prefer, as it's intended for embeddings. Essentially, what node can serve as a replacement for Redis?

You might want to explore the “chat memory manager” and “window buffer memory” (found under Advanced AI > Other AI nodes > Miscellaneous > Chat memory manager) and connect it to your AI agent.

If you're open to a more intensive solution like using the in-memory vector store (though I'd advise against it!), another option is to paste the spreadsheet data directly into a code node and utilize it that way.

Thanks again - this should also work for self-hosted options

This discussion was automatically closed 7 days following the last response. New replies are no longer permitted.