Is there a way to automate task creations using the summary function in Front?

We've done something similar with external AI agents. We utilize a webhook with a rule. The rule fires, sends the webhook to the AI agent and then it processes and either generates a reply or comments via API on the conversation with the relevant information. This approach allows you to train the AI and give it all the needed information to accurately respond to your specific industry/company needs. 

This is a really good idea — and it’s something we’d love to see.

We already use Tasks, but it would be extremely useful if completed tasks automatically showed a checkmark inside the original conversation. Even better, if completing the task automatically re-opened the original conversation, it would notify the original assignee and help prevent anything from slipping through the cracks.

It would also be great to have a simple progress-bar style view showing all tasks linked to the conversation, so we can quickly see what’s pending, what’s completed, and how far along the workflow is.

​@broc_afseth Sorry for the late reply. Really interesting use case you’re sharing. I’d like to dig in more to make sure I understand.

• With this automatic task generator, are you leaving it up to AI to basically create a list of how it might solve the customer problem? Is there any additional guidance or knowledge the AI needs to have to do this?
• I’m not fully following this part: “This way too, if a customer requests what a process looks like, then we can give them bullet points and a look into what our process was” - Do you expect to share those tasks with the customer if they ask for a status update, for example?
• Do you have specific operating procedures (SOPs) that should typically be followed when responding to customers today? This seems like the more predictable and standardized way to handle tasks. For example, if a customer writes in about a topic, and a rule in Front automatically creates a set of tasks specific to that topic. 

On occasion, when working with new personnel who are taking over old CAMs positions, or new customers entering our network sometimes helping them understand how we work our processes and what our limitations are, helps eliminate confusion and streamlines communications. 

     Yes, we do have SOPS for certain tasks and functions, but not for all of our actions, some of them require us to use deductive reasoning and hands-on investigations to confirm differences in cases and pallets that functions and synapse itself may not have correct. That's why some modifications would need to be required. 

The AI would generate a general synopsis that we can expand or condense what’s needed; each bullet point that the summary creates can then be generated into a task, i.e., summary: customer is requesting a confirmation of TIHI and is requesting what the process would be for updating the TIHI.  

AI generates a blank task format using the request, “TIHI confirmation”. 
     Using the SOP, AI then generates a task checklist based on steps used to do a TIHI check. We can then add or remove steps in order to streamline the process for the clerk. Secondly, we can then copy this task list and respond to the customer, letting them know this is what the process looks like and what to expect when conducting a TIHI check.  as the checklist is completed and the steps are met, a progress bar is filled e.x, if there’s 5 points in a check list to complete, and the clerk has completed, 2 of them, then the progress bar would show, 40% complete. 

    Using the SOP as a source for how the AI formulates the tasks would ensure that we follow SOPs which makes sure we are consistent in how we handle tasks and reduce the risk of COC failures. 

    As mentioned above, by Gareth, we can then see a progress bar based on the clerks' responses and filled checklist, which we can forward to a customer to provide visual representations of how much is being completed for large task like De-picks and PIs, and what remains. Which should reduce the amount of “following-up” emails, as the customer and all involved see live progress on the task.