You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
We'd like to explore an Apollo adaptor/integration so that users can call out to specific services from workflows.
For me, the smaller, more controlled and more focused those services, the better.
So for example: filtering emails or grievences.
We could have an Apollo function like filter(data, criteria) to use AI to filter an array of items, orgroup(data, criteria, groups) to group an array into an object of arrays sorted by key.
Importantly, this stuff is good at inherently fuzzy searches. "Filter grievences by urgent ones" or "attach the best label to these emails". Not for "filter items where x is greater than 10" or "sort by this key".
This is run-time AI usage and absolutely has considerations from a responsible AI point of view.
Consider:
how API keys are set and passed (I guess that's just on the apollo credential?). Or, who pays for this? Is it part of your project's AI credits? Actually yes
whether the model can be swapped out
What models we might provide for this (I want to say the smaller the better I fear this isn't true)
The text was updated successfully, but these errors were encountered:
We'd like to explore an Apollo adaptor/integration so that users can call out to specific services from workflows.
For me, the smaller, more controlled and more focused those services, the better.
So for example: filtering emails or grievences.
We could have an Apollo function like
filter(data, criteria)
to use AI to filter an array of items, orgroup(data, criteria, groups)
to group an array into an object of arrays sorted by key.Importantly, this stuff is good at inherently fuzzy searches. "Filter grievences by urgent ones" or "attach the best label to these emails". Not for "filter items where x is greater than 10" or "sort by this key".
This is run-time AI usage and absolutely has considerations from a responsible AI point of view.
Consider:
The text was updated successfully, but these errors were encountered: