In June I gave a presentation at the O’Reilly AI Conference about the operations effort that enables Clara to combine people with machine learning — something we call Cooperative Intelligence. You can find the talk at https://youtu.be/rf2asCOOvfQ, and the slides at here (hi res available here).

 

The main thesis of this talk is that work can be broken down into small tasks of varying complexity and we can progressively achieve speed and cost gains by matching tasks with the agent best suited to process them. In some cases this agent may be only a machine, but in many cases depending on how much of a task can be predicted by algorithms, this may be different credential-tiers of workers.

 

How we allocate increasingly complex (decreasingly automated) work.

 

This concept is illustrated by the image above. Each work queue represents a tier of increasing complexity (or potentially less automation) and one must be increasingly credentialed to obtain work from higher queues. Work that is fully automated appears in the leftmost queue while work that requires complete manual override appears in the rightmost queue. The queues in between contain a mix of partially automated tasks. Since workers can escalate or pull work from lower queues, the system effectively optimizes task allocation and corrects for when our heuristics fail to pre-sort tasks appropriately.

 

For more tricks and tips for building a system like this, see the talk.

 

Clara is growing

 

We’re excited to grow our platform and team. Find out more about open roles here: https://jobs.lever.co/claralabs

Learn more about our CEO Maran Nelson here:
https://www.forbes.com/sites/clareoconnor/2017/04/18/clara-labs-wants-to-save-your-from-your-inbox-with-cyborg-assistants/

and read about our recent $7M series A announcement here:
https://techcrunch.com/2017/07/19/claraseriesa/