It was just a joke to some when it burst onto the recruitment landscape years ago, but Tengai, the robot recruiter born in Sweden is still here, suckers. And now it speaks English, so the boys thought it was worth a chat about what's going on with the company, the tech, and that craziness that is Swedes eating rotten fish. Yuck! Virus-free, fluent in English and a little less creep than last year, Tengai isn't going anywhere. They don't care about your shade. Sorta like Abba's Greatest Hits.
This episode supported and powered Sovren, the AI, parsing, and matching tech so human you'll want to take it to dinner.
PODCAST TRANSCRIPTION sponsored by:
Sovren (1s):
Sovren is known for providing the world's best and most accurate parsing products. And now based on that technology come Sovren's artificial intelligence matching and scoring software. In fractions of a second receive match results that provide candidate scored by fit to job. And just as importantly, the jumps fit to the candidate make faster and better placements. Find out more about our suite of products today by visiting sovren.com. That's sovren.com. We provide technology that thinks, communicates and collaborates like a human. Sovren Software so human you'll want to take it to dinner.
Tengai (42s):
This is the unbiased interview robot, you're listening to the Chad and Cheese podcast. I love these guys.
INTRO (52s):
Hide your kids! Lock the doors! You're listening to HR’s most dangerous podcast. Chad Sowash and Joel Cheeseman are here to punch the recruiting industry, right where it hurts! Complete with breaking news, brash opinion and loads of snark, buckle up boys and girls, it's time for the Chad and Cheese podcast.
Joel (1m 11s):
Oh shit. It's Monday. And we're talking robots. Bri is back in the house. Everybody, you know, you love them. This is Joel Cheeseman your co-host for the Chad and Cheese podcast joined as always by my faithful sidekick Chad Sowash. Chad, how are ya?
Chad (1m 28s):
This is robot day, man! I can't wait!
Joel (1m 30s):
Robot day. Yeah. Let's let's bring in the duo, straight out of Sweden. Sinisa head of product, Elin, CEO of a Tengai Unbiased. Guys, how you doing?
Elin Öberg Mårtenzon (1m 44s):
It's been awhile. It's been awhile. Yeah, we're doing good, given that 2020 is a really crappy year though, global wise and pandemic wise. But otherwise I think that we're quite good. What do you think Sinisa?
Sinisa Strbac (1m 57s):
Definitely. Definitely. It's going to be fun to talk about Tengai today!
Joel (2m 0s):
He's had too many cups of coffee I believe.
Chad (2m 6s):
Since we're in 2020 and we can't get out and people can't really do the see, feel touch of the recruiting robot. How, how much harder has it been to be able to get people interested and also adopting because before, I mean, once I got in front of the Tengai and started to interact, it was like an automatic sell. Right. But it's harder to do that now. So what's it like today?
Elin Öberg Mårtenzon (2m 33s):
It's challenging for everyone, that's not being able to meet face to face. So we're adjusting as, as, as part of, as part of this situation. Of course. So we're doing a lot of demos online or doing a lot of video meetings and things like that. But I think the, the interest in Tengais is still quite big and we're trying to adjust to them going Tengai over video on the brighter side, I mean, isn't able to get the virus, which is quite good. So using that situation is, might be quite beneficial though.
Chad (3m 7s):
Agreed. So is that, is that like one of the big selling points now you still have these, these essential workers that have to come in, they have to interview, but you want to mitigate risk. Tengai will help you mitigate risk and be unbiased during the interview. Is that part of the pitch now? I mean, because this, it had to have changed.
Elin Öberg Mårtenzon (3m 27s):
Yeah. We're actually, we actually have that as a pitch moving sort of the, the early stage of the pandemic. So we used Tengai together with a municipality in Sweden for, for that specific purpose where they needed to recruit people to their health care. And we use Tengai for that purpose and knowing that we could minimize the risk of contamination during that interview.So yeah! Might be a selling point.
Joel (3m 55s):
So you we're here today primarily because we have something that you guys have been working on for a long time. Obviously you're very excited about it. Our listeners who have been following you for the past few years have been waiting for this English speaking Tengai.
Chad (4m 12s):
Today's the day kids!
Elin Öberg Mårtenzon (4m 15s):
Finally!
Joel (4m 16s):
Should talk about the engineering challenge around that. Cause I think we probably don't have a, an appreciation for probably how challenging it was. You know, we have UK English, we have American English, Australian English, we have different sort of angles and context to all of that stuff. So talk about the unveiling and then the engineering challenge challenges behind it.
Chad (4m 38s):
I want to know if you have Mississippi English, because that's a feature.
Joel (4m 44s):
Hillbilly Tengai coming in 2022!
Sinisa Strbac (4m 48s):
We decided early for American English. Cause actually what we think or that you can maybe feel, feel in that UK is more <something Swedish?> In US is, is not as interesting for Americans as American in is in Britain. So we decided to go for American English and then it was mostly about translating and making it contextualizing questions doc, so that they both sound right and have the, keep the core of the questions, as in Swedish, as in English. So we have been working with a couple of translators, both that are both actually British/American.
Sinisa Strbac (5m 29s):
We have also Australian/English as well. So we were like working cause we, we did like Swinglish first, you know, Swedes translated to English and then we learned, we're like, well, this is, this sounds good. Well, Americans and British listened to it. And we're like, okay, this is, this is not how we, how we talk and how we pronounce stuff. So, so they did a, they have done a lot of, a lot of work on it.
Elin Öberg Mårtenzon (5m 60s):
Yeah. And I think that not only that, it's also a question of, of the back channels, like how we do the, how we do the, uhhuh, eh hmmm, the cultural things that we have in a dialogue, which is different from, from the Swedish way of doing it, and into UK, us and all of the world actually. We do a lot of different things, cultural wise, you know, dialogue that we needed to find a solution for it basically so that candidate can feel safe and secure during their interview.
Sinisa Strbac (6m 30s):
Swedish, Swedish people, When, when w when you were talking, the one that's listening is always like, mm, mm, mm. And we did, they did actually have that in the first English immersion. And most of them that are not, that are native English. We're like, why is it doing that? Why is it important?
Chad (6m 48s):
Can you tell your robot to shut up? I'm trying to answer a question.
Sinisa Strbac (6m 52s):
So, that's gone in English.
Joel (6m 54s):
So how many engineers did it take to build this thing and how many hours?
Sinisa Strbac (6m 58s):
Oh! Engineers. I mean, we had like a couple of translators and then what can four to five months translating and testing, coding, retesting.
Elin Öberg Mårtenzon (7m 12s):
Then the main part of product development is testing to understand, okay. Does this actually work in, in an actual interview? How does, how does the candidate respond? How, how do the candidate interact with the robot?
Joel (7m 27s):
How do you guys do that? Do you bring in random people to do it? Do you use internal
Sinisa Strbac (7m 32s):
First internal then, then definitely PR pretty soon after we do, we bring in external candidates.
Joel (7m 39s):
Okay.
Elin Öberg Mårtenzon (7m 40s):
As it is when, when most technology, the more data the better. So the more interview data that we can can get in order to understand if, if it's a good or a bad interview, basically the better. Yeah.
Chad (7m 52s):
Did you have access to Americans? Are in there in Sweden that you kind of plucked out and put in a room with Tengai? Is that what you did?
Sinisa Strbac (8m 0s):
Yeah. Yeah.
Joel (8m 1s):
So you have the shiny new toy. Talk about the rollout. Is it like European and English speaking companies? Is it worldwide, like, talk about how you guys are targeting and enrolling this thing out?
Elin Öberg Mårtenzon (8m 12s):
In the beginning, we had the plan was to launch it in, in US during beginning of 20 20, and 2021. But given, given the situation, the global situation and the possibility of moving and be physically there, it's the challenge, of course. So the strategy now is to work with Swedish multinational brands that has English as their, one of their needs for expanding this product in their organization. So that's the first, first sort of roll out that we're doing. And it's also been a demand for our clients in Sweden to have the English version as well.
Elin Öberg Mårtenzon (8m 52s):
And in parallel with that, we do have partnership program that we started out in during the spring, just before the summer. Which is, it started out with sort of the corporation together with CNG, who's the partner company in Sweden that we're working with. So they have put all their salesforce within this rollout in Sweden, and they have actually been waiting for this new release that was released 1st of September with an updated software with the English scale and with an updated sort of interview with updated interview concepts. So that is also one part of the role as strategy to go with partnerships.
Elin Öberg Mårtenzon (9m 33s):
So we do have a current sort of dialogues with companies in US and UK, primarily
Chad (9m 40s):
So how do you target partners, and then also brands? Because it's funny, we've talked about Ikea because obviously they are in Sweden, but they also have a huge footprint here in the US as well, and, you know, I would assume across all over the world, how do you target those brands? And are you looking for partners to do that? Are you guys doing that or is it a combination?
Elin Öberg Mårtenzon (10m 3s):
I think it's a combination looking at the targets in, in Sweden, we've been, we've been targeting companies that has really high demands on sustainability, on diversity and inclusion. And the companies that actually want to do the next step of their diversity inclusion efforts, so we've been working quite closely with them.
Chad (10m 26s):
What about you talk about diversity and inclusion, which is obviously big over here in the States. You've, you've noticed that there's some huge movements, which have been long awaited. Unfortunately they had to be sparked off with, with, with some very, very bad incidents. When that happened, it just seemed like it seemed like diversity, equity and inclusion really rose to the top and being unbiased is incredibly important, but there are a ton of people that are out there that are haters and don't believe that any algorithm can be unbiased because to be quite Frank, most algorithms are programmed by white dudes.
Chad (11m 11s):
Right? So what's your, I know there's a science behind it. Can you talk us through, help us understand how this technology, whether it's physical or just the algorithm is not biased?
Elin Öberg Mårtenzon (11m 23s):
We've been doing this for quite many years in Sweden, and we've been doing unbiased recruitment in a manual way for 15 years. And what we have sort of explored, during those years is that a process needs to be data collection, and we need to collect the, the right amounts and the right data in order to make better hiring decisions. So looking at the interview, which, which is Tengai's topic, we need to look at what the interview that we could collect with data that makes it unbiased. And it needs to be structured. It needs to be anonymous. It needs to be focused on the things that can be related to behavioral science.
Elin Öberg Mårtenzon (12m 9s):
So those are the things that we have been focusing on extracting the right amount of data from that. So we don't use, for example, we don't use data such as age or gender or ethnicity or race or whatever. We only use the behavioral science data that doesn't have a discriminatory round. And from that, we've actually done a rule-based situation pretty much as used in personality tests. We know that there's, there's a given fact when it comes to performance in different areas in personality. So we've actually just put those things together with the parts from the personality tests and the parts from the behavioral science and into what we call an interview hybrid.
Elin Öberg Mårtenzon (12m 55s):
So we're combining those two things. And perhaps initially you can explain how the personality indicator and the behavior indicator actually works.
Sinisa Strbac (13m 4s):
The TPI is like overall name and TPI is Tengai Performance Indicator. And that's both like the framework, but also the score that every candidate gets at the end of the interview and the TPI, is done a combined, is it actually a combination of two different scores, as Elin mentioned, it's personality indicator. And then you have the behavior indicator, personality indicator is statements that the robot asks a candidate and that candidate needs to either agree or disagree, but either by saying yes or no, or, or giving like longer answer that interprets into either positive or negative.
Sinisa Strbac (13m 47s):
And this is connected to studies about big five and general work performance. So it's not, it's not like a whole big five, but part of big five that is being measured by this first like personality indicator. And that score is actually out dramatic in this new version. So right after the candidates complete the interview, the score is already in the software. So recruiter need and knows already that, okay, this candidate is good to go move forward or not.
Elin Öberg Mårtenzon (14m 20s):
Yeah. Th that's also the part that that's been validated that you announced earlier this year. Yeah.
Chad (14m 26s):
Well, here's a quick question around validation. And also there's a huge population of individuals out there who don't act and or respond the same way that, you know, Bob does. And those are individuals with disabilities. Did you guys do advanced testing to ensure that every different segment was represented in testing to ensure that there truly wasn't any, any bias even in that sector?
Sinisa Strbac (14m 53s):
Yeah. Well, I mean, we did, we had done some tests and as far as we know, 98% of the answers that Tengai has interpreted has have been right. I mean, the candidate has, has talked, Tengai interpret this was it, this was positive negative, and has like 98%
Elin Öberg Mårtenzon (15m 13s):
When it comes to people with disabilities is the, for example, we did a lot of testing on people with autism, had a little difficulties in normal human human interviews, since they had problems, keeping eye contact, et cetera, and they needed more time to respond, et cetera. And that worked really, really well with this type of technology. So they felt more positive after the robot interview than they felt after a human to human interview, simply because the robots didn't fill in gaps or just waited for, for the candidate to actually give their view of things. So from that perspective, that was really, really good.
Joel (15m 52s):
I'm going to address the elephant in the room, real quickly. You guys are, I'm going to say four years old at this point, somewhere around there. And I would, I was one of the first that sort of reported on you guys. And I would say the, the comments I generally got was like, this is a PR stunt. This is a joke. These guys will be dead in six months and those comments still continue today. But here we are, four years later, you guys have launched an English for anyone who thinks, you know, this is still a joke. This is just this isn't going to happen. Robot recruiting is not going to be a thing. What would you tell those folks?
Elin Öberg Mårtenzon (16m 28s):
Well, I think that, you know, most people that are very negative towards us and think that it's a gimmick or whatever, they haven't actually met Tengai and they have never sat down in a room with Tengai conducting this sort of interview. And we know from experience that people that has been negative in the beginning, sitting down with Tengai, they have a different approach afterwards. This things that they most always tell us, is that, okay, this was not what I expected. This was a really serious interview. It was really hard, but I also felt that I could be sort of evaluated for my skillsets and nothing else. I totally get the purpose. I look at this as more of a data collection tool than anything else than more than a robot,
Joel (17m 12s):
Are you willing to talk about how many customers you have or is it not even necessarily about that as much as it is, we're still building the product?
Elin Öberg Mårtenzon (17m 19s):
I think that we're still building the product and we're mainly focusing at the moment locally. We do have a few clients in Sweden that we're super happy about. Still within municipalities, a few banks, were having a really good dialogue together with Swedish authorities, et cetera. So, but, but still for us, it's, it's been really important to build a really good product and to, to focus on the science part of it and the, to as so basically when we launched the first product, 1st of September, which is the combination between the automatic and the manual part. Yeah.
Sinisa Strbac (17m 55s):
Yeah. I didn't tell you about the second, the second part. So just so, just so you know.
Elin Öberg Mårtenzon (18m 0s):
Yeah. Give us the behavior indicators in a chat. Yeah.
Sinisa Strbac (18m 3s):
Sorry. So, so the first score is automatic and that's based on personality of the candidate related to general work performance. So you get like the whole, the group, the group of candidates, they get interviewed, you have the first score automatic, then you have the second, the second part of it is it's called behavior indicator. And that's like a past behavior questions. Like how do you solve a problem? Given that situation where you did let, what did you learn? Some, some, something similar like that. And that's where Tengai records answer and then presents the answer to the recruiter, like audio files that sends it, the recruiter can score candidates according to a grading system that we have created.
Sinisa Strbac (18m 53s):
So you have like, the robot does the first scoring where candidates states, that I am like this, this is my personality, I'm good for this role. And then you have the audio scoring that we call it where the candidate also gets to describe it's competence. And that's like, where, what we are thinking that magic happens when you have like, okay, candidates says he, or she is like this. And they can also describe when you have these two, like, okay, probably it's a good, it's a good candidate to move forward with.
Joel (19m 26s):
And you talked about, sorry, you talked about sort of the, what I'm S I'm assuming what employers think when they meet Tengai and go through the process. What has been the reaction with job seekers as you guys roll this out? And is it what you expected or is it different than what you anticipate?
Elin Öberg Mårtenzon (19m 44s):
It's both what we expected and some new things has come up along the way, but what we mostly get when the, sort of the, when the shock with the technology sort of lands throughout the interview, because that's basically what happens. The first thing, when a candidate hadn't met a robot before, and they're starting to get sort of intrigued and then feels like, okay, this is amazing, et cetera. But I think that when that has sort of, sort of landed a bit, they feel like, okay, now I have started my interview and I'm talking to this white piece of plastic, but all of a sudden they start to forget that it's actually a robot that, that they're talking to. And that is what we felt as well when we started to work with this product and this platform, because it's so interactive and it behaves just like a human and it's natural for us to actually feel that it's human communication that we're sort of doing.
Elin Öberg Mårtenzon (20m 37s):
So, even though we can see it's a white piece of plastic and it's that it's a robots, the brain sort of interpreted the communication as human. So the brain tells us that it's a human that we're talking to. So that's why sort of creates this strange feeling for people. Okay. I can see it's a robot, but I felt that it was, I sort of forgot that it was a robot I was talking to. And that's natural, I think.
Chad (21m 3s):
To say that I was definitely one of those haters at first, before we went to Portugal and actually met Tengai, I thought, this is, this, can't be real. The interaction can't be there. And it was, and you said something earlier was eye contact. Tengai makes eye contact, makes those subtle gestures. And it was really interesting just to engage. But that being said, knowing Tengai can make that eye contact and is looking for that eye contact. We're hearing some, some pretty bad legislation coming out for facial recognition. Does Tengai do facial recognition? And if not, will she moving foward?
Elin Öberg Mårtenzon (21m 44s):
She will not use facial recognition. We do use face tracking, which yeah. So, and it's the purpose of that is basically to make sure that the robot can follow the face that it's tracking. So it's more of an interactive experience for the candidate. There's also quite many things coming up, but when it comes to face tracking such as smile detection, et cetera, as Sinsia as you can tell, tell us more about that, perhaps, but that's, we will never, and we don't have that in our, in our pipeline to use facial recognition in terms of connecting that to different behaviors.
Applause (22m 26s):
Clapping.
Joel (22m 27s):
Well guys, as always, it's been a pleasure for our listeners who don't know about Tengai. Where do you send them?
Elin Öberg Mårtenzon (22m 36s):
To tengai-unbiased.com.
Joel (22m 38s):
You better spell it for people. Yeah.
Chad (22m 42s):
Tengai see I even screwed it up. Tengai-unbiased.com guys. Thanks a lot. And Chad We Out. We out
Joel (22m 51s):
How do you say we out in Swedish?
Sinisa Strbac (22m 53s):
Vi ut.
Chad (22m 55s):
We out.
OUTRO (22m 56s):
Thank you for listen to podcasts with Chad and Cheese. Brilliant! They talk about recruiting. They talk about technology, but most of all, they talk about nothing. Anyhoo, be sure to subscribe today on iTunes, Spotify, Google Play, or wherever you listen to your podcasts. We out.