The client, us! Had a problem. We were growing fast and needed lots of people to join us, but, we had a specific way we like to recruit.
You see, although we have two offices, we don't demand staff work from them. We recruit on a global scale. We also don't like being specific on what skills we want. We believe to get A-players you have to cast your net wide and, rather than saying "We need X years Tensor, X years machine learning and a degree from Stamford", we say "We need geeks, let's chat". If you get specific, then only the people who meet those exact requirements will make contact. What if there was a machine learning genius who didn't know Tensorflow? She would never bother applying...
The problem is, when you shout to the world that you are hiring geeks, a lot of geeks shout back.
Now, for the non-AI company, this would be a problem. Thousands upon thousands of people from around the world making contact. Who has the time to speak, on a personal level, to all of them?
Time-pressed, conversational mass-volume of input? Sounds like a job for our geeks!
Turning recruitment into an RPA-chatbot led process
We built ubibot.
It's an RPA/chatbot that deals with inbound employment enquiries. We tell ubibot the types of roles we're looking for, a few qualifiers that we need for that role, and it sits there all day, across multiple channels, speaking to people about those roles.
When people answer the few questions, ubibot saves their details and sends a follow-up email to confirm it got everything correct (and perhaps to check their email address wasn't a fakey ;) ).
Now, that's all pretty bog-standard chatbot stuff. However, here comes the magic. ubibot doesn't tell its humans about every application. It knows what we're looking for and makes an intelligent decision when it gets one we think we should know about. Take the example above; we want Tensorflow and machine learning experience. Well, if ubibot gets someone who's a PhD in ML but hasn't specified Tensorflow - it thinks we should know about them.
ubibot presents prospects to us, and the humans tell it whether to proceed. ubibot sends emails, organises the first contact between the humans and collects responses along the way. As a candidate is progressing through our recruitment process, ubibot is messaging them and keeping them up-to-date on where they are and how they're getting on (and asking for clarification or more details as required). After all, no one likes being left in the dark right?
It's also pretty smart at machine learning stuff. When we have a new role to hire, we tell ubibot. As well as talking to prospects about that role, ubibot is actively working in the background and understanding the interrelationships between the different types of data that exist within our CV, candidate, role descriptions and job advertisements.
The future of ubibot
It's always challenging to manage internal vs paid work (after all, we can't just make cool stuff for ourselves all day, we have to pay the bills too!). Although ubibot is impressive and saving us hundreds of hours (while nabbing the best talent!), we have a long roadmap of features we're adding.
Here's a headline summary to whet your appetite:
When we produce a job advertisement, ubibot will guide us and make recommendations based on previous successful advertisements and external data sources.
When the advertisement is ready to be published, ubibot will suggest appropriate platforms and budget for publishing based on previous success metrics, industry, role, target candidate etc.
ubibot may research other publicly available information about the candidate from external sources, such as LinkedIn (a bit scary so not too sure about this one).
When a new job role is required, ubibot will reach out to people in our CV database and see if they're still looking for employment and, if they are, update their details and tell them about the new role.
Finally, we've got some crazy idea about turning ubibot into a digital interviewer (like, give it an avatar and stuff!) and using image/audio/video capture for analysis and recommendation creation - although, the technology to make it production ready is still quite far away!
They're just the features we're building for ubibot to be more useful for us.
We haven't even touched on what we could do for recruitment agencies and enterprise clients who are dealing with tens of thousands of CVs and candidates...