People-centric question answering

People-centric question answering
Photo by Adi Goldstein / Unsplash

Consider for a moment, you are at work and you have a question. Maybe you have an expense that you aren't sure how to submit for reimbursement. Perhaps you want to know how to get access to a new piece of software. I asked my wife, an environmental scientist consultant, for a recent example - "Which contract template do I use for a client change order?". You have two options - you might ask a coworker who you think might know or you might search to see if someone has written something about this.

Connie AI can save you and your coworkers time by greatly improving the searching side. There is a problem though. Connie can only know what has been written down. We aren't aware of companies where everything someone might ask about has been written down, and even when something is put in writing its just as hard to keep it up to date. The term "living document" is well-meaning but rarely accurate in the long run.

We want Connie to always be helpful when asked a question. We realized that Connie can't do that unless it understands the people behind the documents as well as it understands the documents themselves. Let's look at some examples:

Missing knowledge

When Connie is unable to find an answer to a question in your Confluence space, it doesn't stop there and give up, as that would leave you at a dead end. Instead Connie uses its knowledge of the documents in the space and considers who has written about topics that are similar to the question at hand. Connie then ranks those individuals and presents to you some people you might ask so you aren't left at a dead end.

Outdated knowledge

While Connie is aware of the temporal aspect of your documents (when were they authored, updated, and by whom), Connie usually can't know that a particular piece of writing is no longer accurate when a question is asked. In this case Connie will provide you with the answer it found, but again will make use of its understanding of semantic similarity and author contributions to suggest the most likely experts on that particular question. That means if you find the answer to have been outdated, or even if you simple want to dive deeper on the topic, help is just a quick chat message or email away.

This is just the start

Connie is still a brand new product that we are actively refining and experimenting with. We think that placing a stronger emphasis on the people behind the answers holds a lot of value. We have our own ideas of how Connie might continue to unlock new value here, but we'd love to hear from you. Would this be useful to your company? What features might you add to be more helpful? Send us an email to as we'd love to have early adopters help to guide Connie's direction.

If you'd like to be notified of new posts and announcements from us about Connie AI and future products you can subscribe below: