Why We Wrote a Babajob Chat Bot
Babajob’s mission — Better Jobs for Everyone — is made possible by enabling employers to digitally and reliably hire aspiring workers. This mission has motivated and driven my life and many of my colleagues who have built this company for the last decade.
New technologies are created constantly and one of the things that makes software so special is that as a developer, you can use those innovations and build new innovations on top of them remarkably quickly. This process of leveraging previous innovations and adding to them is how the world spreads productivity. Open source software and the history of science are based on building on the ideas of others but so are companies like UBER and OLA who could not exist without platform innovations such as credit card payments, ubiquitous smartphones with easy to integrate maps and location.
At Babajob, we are trying to solve a few fundamental hiring problems through technology. For example, imagine you are the hiring manager for a pizza delivery company. You have a recruitment center outside of a city. If a male between 18 and 25 years old shows up at 11am on Wednesday, with a valid motorcycle license, 10th pass certificate, knows basic English and wants to deliver pizza for 8 hours / day for Rs.12000 / month, he has a 90%+ chance of walking out with a job offer in hand.
Hence, Babajob’s purpose as a platform is to get everyone that’s interested and qualified for a better aspiring job to apply and then show up for interviews with employers that want to hire them.
We break this problem down into a few key parts:
- Enable job seekers to digitally represent themselves well, by using just their smartphone, but do so on top of the spotty network infrastructure of India, where folks may have great, free Wifi at the Metro station but only intermittent or non-existent 2G on the bus or at their home.
- Provide interfaces in local languages given that many of our job seekers may not read in English well.
- Solve job seeker verification and data collection challenges, cheaply and at scale, including:
- Location: knowing where someone lives (so that we can tell them about nearby jobs).
- Verifying their government and school documents.
- Assessing their communication and language skills/
- And now — What kind of personality do they have? Extrovert vs Introvert? What types of jobs suit that personality?
- Safely connect employers and job seekers together over the phone — without revealing phone numbers — and facilitate interview show-ups.
Recently, we’ve seen several new innovations that we believe can address the above problems:
- SMS and now Chat are now the most commonly used digital communication tools; WhatsApp and Facebook Messenger have become cheap, reliable ways to communicate even in poor network conditions. The fact that Facebook also offers photos of their users and a degree of verification (people usually don’t lie about their age and name to their friends) is helpful too, given we can reuse that data for employers.
- Cognitive APIs such as text analysis, speech and image recognition have become commodity developer APIs (thanks Microsoft, Facebook and Google!). For Babajob, this means that with a moderate amount of work, we can analyze a photo and assess whether it’s a picture of a job seeker’s PAN Tax Card or analyze her speech to assess her English level. Soon, we should be able to analyze most chat messages from job seekers and provide a reasonable answer back in their local language, be it Hindi, Tamil, Kannada with more to come soon.
- India Stack: India has made some incredibly public policy investments with Aadhaar — which is akin to the US’s Social Security numbers, but with verification. That means that anyone verify that an Aadhaar number, name and registered mobile number match what’s on the government records via APIs. This enables businesses like ours to verify that people are representing their identify truthfully and importantly gives us, employers and job seekers recourse with the authorities if something goes wrong.
Here comes the bots!
Given all this, we felt there was an incredible opportunity to combine these industry trends into a better way for job seekers to get assessed, verified and hired.
We’ve built a chatbot available on our website and the Babajob Facebook page which has over 500,000 fans who are looking for jobs. It enables job seekers to simply search for jobs and build a great profile to showcase themselves to employers.
It offers a few innovative, never-been-seen-before features:
- Search Jobs in 3 chat steps: we ask for your mobile, location and category and immediately show matching jobs.
- A Personality Assessment Test, created by Babajob and the World Bank, enabling job seekers to showcase their best personality traits.
- Seekers can snap their resume, Driver’s license, Aadhaar, PAN Card, etc and we’ll decipher their info, verify it with the government and showcase their verified documents to employers they apply to.
- Send a voice clip over Messenger and we assess a jobseeker’s English skills (and dialect) and showcase their voice clip and English score to employers.
- A Private Phone Number for every job. We protect the privacy of employers by giving a virtual number to jobseekers after they apply. When job seekers call, they are first asked the employer’s specific screening questions and only thus seekers that pass the screening criteria are call connected to employers.
- All this available in English, Hindi, Tamil and Kannada
We make all this possible by leveraging the latest cognitive and AI APIs from Facebook, Microsoft and Google. We also use India Stack verification APIs to verify Aadhaar and PAN Cards. For the geeky folks out there, here’s the API diagram.
In conclusion, we are terribly excited about what we’ve made and as the bot’s primary developer, it’s been very fun for me to get deeply connected with code again.
Cheers,
Sean Blagsvedt
CEO, Babajob.com
PS. Given the Quikr acquisition, we’ve working on redoing the bot against the Quikr APIs in the coming months. In the meantime, you can get a sense of the experience with the video below:
