About Jobby
It’s the day before going on my holidays for two weeks. My luggage is packed, the house is cleaned, everything is ready. Looking around for something to do, suddenly remember a project I have worked on a while ago (almost 6 months ago), it was called Jobby a chatbot to interact with recruiters.
Motivation 👨💻
The idea started when I wanted to switch from permanent to contractor. Seeing how many messages I was receiving on LinkedIn and how difficult was to comb through them, I knew I had to find a better way.
Messages with job opportunities on the wrong technologies/stack, or offering salaries at a junior level, as well as other factors, made it frustrating to read every day. I wanted to have a database of potential roles that I could filter, organize and sort by different criteria.
The tricky part was to find how to enter all this data into a database. The first idea I explored was to parse all the emails that came from LinkedIn, and use some kind of machine learning classifier to extract the information I was interested in. After collecting some messages and trying to do it “by hand” I quickly realized that recruiters rarely give me the information I needed in the first message. They usually don’t tell you the name company the role is for — they say things like Blue Chip company, Award Winning X, or Leader on its field — additionally they rarely give you the salary, a range at most is typically provided instead.
So my next thought went to chatbots, I have been playing with chatbots for a while, having made a simple Slack bot and trying different AI and NLP frameworks. I was really impressed by the new Bot Framework by Microsoft, it allows you to create dialogs easily and publish your bot on different platforms. It allowed me to ask for the information I wanted from recruiters, and create a database of job opportunities I could filter and sort.
Release 🚀
Fast forward to the day before my holidays again, I open the project and it was almost completely finished, it was deployed for free on OpenShift with a MongoDB cartridge ✓, Facebook bot created ✓, Skype bot created ✓, and with a dialog to get information about contract roles ✓.
Having decided not long ago that I didn’t want switch to contract anymore I had to change the dialog for both permanent and contract — I did forget a question about duration on permanent which many people pointed rightly out that didn’t belong there.
And so with just an afternoon to get everything ready, I published an improvised post on LinkedIn basically telling recruiters I would only consider roles coming through Jobby. I thought that would be the only way this could work, otherwise it would hardly make an impact.
That post was supposed to be a first test and my expectations were pretty low. I hopped that a couple recruiters might use it and I could get some feedback to make it better in the future.
Viral explosion 💥
I heavily underestimated the viral potential of LinkedIn. My assumption was that my post would only be distributed to people on my network, but I quickly realized that people liking or commenting on it would distribute it on their network and so on.
On the second day of my holiday I casually checked to see how the post was doing and to my surprise it had something in the region of 1.096 views and 4 comments. My thoughts were that it was a success already!
As I’m writing this I am now on the fourth day of my holiday and the post has already over 20.500 views, 25 comments, 103 likes and continues to rapidly increase. I agree its a drop on the ocean as viral things go, but it has exploded way beyond my expectations, with very interesting feedback so far.
Get Marti Planellas’s stories in your inbox
Join Medium for free to get updates from this writer.
I’m writing this article, because I feel a lot of people took this experiment as something I wanted to do against recruiters or as if I’m disregarding the role they play. I’m not. As I explain here this is a tool for me, to help me manage all the information coming my way. This tool would accomplish nothing without recruiters and I want to make this clear.
Having said this, there was another underlying motivation. I really hate talking over the phone. And recruiters have a very strong preference on talking over the phone as a main form of communication. At any time. For any reason. Even if you don’t know them at all. So this tool would also allow me to get the information I need — as a first contact — without having to talk on the phone.
Results 📈
I know its probably too early to share any results, and I will probably add some edits with new data on the future, but I thought it might be interesting to have a quick peak:
- Jobby has been alive for around 5 days
- Has received 80 job opportunities
- Of which 12 were real opportunities from real recruiters
- 5 of them were hilarious albeit fake
I have to admit that I did get more fakes than what I had expected I would, but they were easily filterable. Only looking at the email address would be enough to identify them.
One way to prevent this might be to add a “Are you here to test/play with Jobby?” question since I have noticed a lot of people just wanted to take it for a spin instead of having a real job opportunity in hand.
To summarize; I was very happily surprised of the amount of interest and positive feedback that I have had thus far. I am more motivated now to improve it into something useful for other people as well (when I’m back from my holiday!).
You can use Jobby yourself on one of these platforms:
- On Facebook Messenger
- On Skype
- Or via web
PS: I will answer all genuine job opportunities sent through Jobby when I’m back from my holidays!
The source code for Jobby is available here: https://github.com/beldar/jobby
