Converse AI by Smartsheet

Human Escalation/Fallback

Converse is able to assist in a wide variety of use cases, but we know for a large number of these it is about augmenting and assisting existing people handling requests, rather than replacing them with total automation.

Even if total automation is the eventual goal, it may well take time depending on the complexity of the existing task being taken over, and this is where Converse excels with its ability to have a human/machine hybrid system.

Examples include:

  • Dealing with common support issues on social media, but leaving unknown enquiries for staff members.
  • Immediate engagement with potential e-commerce purchasers with the ability to fall over to a staff member as required.
  • Semi automation of appointment booking services for small businesses, with the ability for them to still directly engage with their users.

Enabling the human/machine hybrid model
This is why we have built a series of features to assist with this:

Triggers/Alerts when requiring attention
Firstly, if trying to increase automation or streamline existing processes, even if the vast majority is still being done by a human, it doesn’t make sense for a human to have to check every message received if it doesn’t require input. So Converse can automatically alert a provider when it doesn’t know how to handle a request or comment from a user as required.

Ability to specify or update contexts and re-enable automation.
If a request comes into Converse and it is unable to identify or recognise it, all is not lost, a staff member can intervene, and rather than have to take over the entire request, can redirect Converse appropriately, specifying the intent and any known entities as required. Converse will then carry on accordingly without needing more involvement from the staff member. Equally if for some reason Converse is unable to parse an individual comment from a user, within an existing conversation, Converse can alert a staff member who can parse it themselves, and then put Converse back to work.

Ability to easily comment/engage as required.
None of the above functionality stops a member of staff joining in or taking over the conversation at any point, and the enduser does not see any difference in how the conversation is presented or updating, giving a seamless process.

Integration into backend ticketing and support systems.
Of course all of this is unhelpful without integration into a companies existing systems, so we are busy building out integrations into third party ticket/support systems to enable this, Intercom and Smooch are already available, with many more to come.

Updated about a year ago

Human Escalation/Fallback

Suggested Edits are limited on API Reference Pages

You can only suggest edits to Markdown body content, but not to the API spec.