Workable CEO Nikos Moraitakis stopped by WorkTech to talk about recent product announcements with WorkTech Founder George LaRocque as part of WorkTech’s “Meet the Work Tech AI Innovators” series. Workable’s recent announcement extends its platform beyond its core hiring capabilities and into onboarding and talent management. They also announced their “AI Copilot,” which is made up of several new capabilities leveraging generative AI and large language models. Workable is an HR platform for small and medium businesses that has powered more than 2 million hires across over 27,000 companies.
Nikos discusses Workable’s approach to AI, augmenting the recruitment, sourcing, and engagement of candidates. He reviews the product capabilities and gives pointers for any HR or Talent leaders that are considering AI in any of their workflows.
The Work Tech category continues to outperform tech at large for global VC investment. WorkTech Market Insiders can access insights and analysis of all of the deals in our Q1 2023 report with a free membership.
Find the full transcript of the conversation below.
Enjoy the discussion!
Full Transcript (Transcript creation is automated, while it may be largely accurate we apologize in advance for any errors)
George LaRocque: Hi, everybody. It’s George Laroque. Welcome back to work, tech. We are continuing with our conversations where we’re meeting the Work Tech AI innovators. And I’m really excited to have Nicos Moratakis, the CEO of Workable, with us. Workable is a global brand. It’s a hiring platform that’s focused on the SMB space. And so I’m really looking forward to learn more on this. Nicos, thank you for being here. Welcome.
Nikos Moraitakis: Hi George, and thanks for having me. It’s a great pleasure to be with you here.
George LaRocque: Great. And why don’t we start with, if you can, tell everyone a little bit about yourself and a little bit about Workable and and then we’ll go from there.
Nikos Moraitakis: Great. I’m Nicholas, I’m one of the two founders and the CEO of Workable. Workable, as you said, is a hiring platform. And actually, very recently we announced that we’re also going into talent management and doing the rest of the HR stuff with Workable. But what most people know us for so far is we are a very popular SMB tool for finding and hiring talent. Just to give you a sense, it’s a ten year old company, rather mature. More than 220,000,000 people have applied for a job to some employer through Workable, and nearly 2 million hires have happened so far. And looking forward to a discussion today. For many years, Workable has been very much focused due to the nature of our customer SMBs to Automating, and a lot of the tasks in recruiting. And obviously the latest surge of AI improvements in technology has accelerated the space.
George LaRocque: Right, well, and that’s a great launching point. As you mentioned, there was an announcement recently, and there was a lot in that announcement. There was the talent management aspects of it and the AI aspects. Do you want to talk a little bit about what those details were?
Nikos Moraitakis: Absolutely, thank you. Yes, indeed. A couple of weeks ago, we released two big things. One is an HRIS component to Workable, which basically extended to be a system that can cater to the entire HR department, not merely the recruiting aspect of it. That is something that obviously in that category of our customers are typically a few dozen or a few hundred employees. So there is a big incentive internally to try to do these things with one platform. From the data standpoint, from the facility standpoint, and this is out already, quite a few, about 20% of our customers have started adopting it so soon. We are coming with timeTAKING out, which obviously is the most important element of the application. This is where we expect the most adoption. So, yeah, we’re excited about this. A very new thing, but it was somehow a little bit of resado by how magical and fancy are the AI stuff, to be honest with you. Look at Workable, we’ve had AI features in production for several years now, and we’ve had the data science teams in recruiting, even starting from NLP, from resume passing and the matching, et cetera. We had a lot of stuff like that for the past seven years. But what has happened right now is that essentially a lot more things, the quality of the outputs you can get with some of these next generation Lambs is that before we had a thing, for example, that made your interview questions, they were okay, but not great. And sometimes you had to fix them a lot so you could see it was there. Now, what happened recently is that suddenly a lot of that stuff with the right tuning and the right data can be truly game changing. So what we did over the past six or seven months is we said, we’re going to package interventions in the software in every aspect of the workflow that brings those capabilities and makes every part of it a little bit better. The feeling is that of a copilot, someone who’s next to you and says it’s of description. I’ll write you one. You want it, friendlier. You want to change this. And then it says, Let me write the interview kit for you, knowing the job description, knowing what you’ve done before I’ll let you edit it. Alton sky keep notes for the interviews for you so you have everything. I’ll organize it according to your interview. It’s let you compare this summarize. I summarize resumes for you. So along the way it goes in every step. I’ll find you candidates. I’ll look at your past candidates and see if you have anything relevant to what you’re doing right now. Perhaps we could even go and show it a little bit later. But essentially, the announcement we did is a copilot that stands with you in every step of the recruiting process and does some of the more trivial and mundane tasks that probably you didn’t want to do or just want to supervise. And it’s not just an announcement. A lot of these things are delivered. People can go today, right? Even a free trial has it seriously. We have it even on the free and generate job descriptions this way and see how it works. And the rest of it is going to be in the hands of user before the end of the year. In the next few months, we’re coming out. Like every month, there’s a new release. So, yeah, that’s where we are. I’m excited to see the usage and to see what happens with those things.
George LaRocque: Yeah, that’s an exciting time. And there are a few interesting things that you just said. One is really about the business. I have to tell you that I didn’t realize how mature workable was. I think probably at about the time that you were starting to work with AI seven years ago. That’s where your brand started to appear on my radar screen. And I think that’s very US view. I mean, I work globally, but you started to appear in the market. As I’m looking at research and talking to users. The other thing is the fact that you’ve been working with AI for that period of time, that kind of legacy with AI or time with AI gives you a really great foundation to look at what’s coming and what’s coming at us so quickly and understand where to use it. I think buyers and folks looking at this technology, it’s important to find partners that aren’t just bringing in the next shiny object, if you will. And I love the name the copilot. What that suggests is the approach of augmenting and taking over tasks where you see fit. But there’s a pilot sort of setting the course. So I really love that. Can we see some of it? Can we take a look at what you’ve got?
Nikos Moraitakis: Absolutely. So let me just go here and share my screen. I have a little video here that I’m going to share. The video has no sound, so I’m going to speak over it. Can you see it now? Is it visible?
George LaRocque: Yeah, it’s there.
Nikos Moraitakis: Perfect. Okay, so basically this is about like three, four minutes, and you’re going to see a workflow of every feature along the way. I’m going to speak through it to explain how it works. So this is for people who have never used workable, the main screen of workable. This is a small company with one job. We’re going to go and create a job. And you see here, this is the job editor, where okay, these are the AI features of workup. We’re giving you feedback on how to create a job, salary that you should set or expect. But here’s the generative feature, the new one, the one that is live and you can drive even today, where basically it takes into account what it knows about the job. Even your company, you’ll notice it knows things about your company and drafts you the job descriptions. Today, 16% of the jobs created in workable by users are made with this already. And it’s rising. It was 12% last week. Now here’s another thing. You posted your job. Now you want to find some candidates. This is an automatic source with explainability. It even tells you if you notice here why it picked the candidates, where it looks in the database of 400 million that workable has. And then you will hear in the new step we’re going to the candidates we selected, we want to send them an email. We’re going to do outlets. It’s going to do personalized outlets and write the emails. Allows you to change the tone later, to automate this in bulk, et cetera. So you see in every step of the way, there are small tasks like that that take a lot of time, that are micro decisions, maybe low level decisions, maybe no decisions at all. And we can help here. We’re uploading a resume. The resume is up. You see the profile summary that AI has created. Automatically you get the resume, but you also get the summary structured. It has even a little bit of an assessment of the fit of their own. So it’s little things like that along the way in the experience. The copilot, as you said, nobody’s doing anything for you, somebody’s just helping you along the way. Somebody who has seen 230,000,000 candidates and has seen 2 million hires and has seen people like you do the same things and understands your company. Now what it’s doing is scheduling an interview using self scheduling. That’s not an AI feature, that’s a feature we had before. See how you blend these things together. But now it’s creating interview questions. It’s going to create an interview kit. You get the idea automatically knows the job. It knows, again, your background, the context is very important here because you can go and do this with GPT right now, but you want it where you are and you want us to be aware of your essentially, this is not just a GPT rapper. There’s prompting on a specialized LLM for this, with examples, with reinforcement learning. And now that people are using it, soon we’ll be able to know maybe is there some cases where they does a better job at the job description, brings better candidates. So it’s going to reinforce learn on its own. Here’s my favorite. No more interview notes. Transcription live as you do your zoom call. And then it associates it with the questions. Then in the evaluation, when you have the scorecard, it can actually position the answers where you need to be aware of them. Basically, it leaves all the decision making to you and does all the stuff where you would be taking notes and not listening to an interview, for example, with so many points and everything. So there’s a few more things further down. There’s even a place where it creates a personalized onboarding plan for the employee when they get hired. Here’s, for example, evaluation, as you can see, you see for its evaluative criteria, the appropriate question is matched by the AI. It’s rephrased. You could do this before, but the rephrasing would be so awkward, that would be useless to you. But now this is the sort of thing we can suddenly do with the context and the data that we cast. So I’m going to stop this here. You got an idea, I believe. Can I just stop saying, all right, that was great.
George LaRocque: I think it’s important you said it earlier, and I’m not sure if it was as we were coming in or whether it was as part of this, but you had said that you can use chat, GPT or these tools to do some of these things. But the important thing is that context and the data. Where am I in the process? What do I know about this job, this company, this candidate, my style, my tone? That’s really important as we’re looking at this. I’ve seen a lot of solutions chasing generative AI. And I really applaud you for going beyond just the fun stuff that we can all do on our own. Write me an email or something and bringing that data in. How are the customers you mentioned the usage is coming up. Are you getting any feedback? How are they responding?
Nikos Moraitakis: Okay. So far, they seem to like it. Let me give you a start. Before the generative option, in creating a job description, about 7% of our users would choose a template okay. Which was the I don’t have anything in my hands option right now. 16% go for generative. Okay. It was 12% last week, so it’s growing very fast. I expect students going to be probably 25, 30% or more now. I’ll come back again in the podcast in a few months to tell you if those jobs got better candidates or more, or if we’re discovering that it’s good at some things or others. Okay. So far, the users seem to be excited to adopt. It one important thing. They like it. They like it and they use it. We notice how many they choose. The Edits Friendly Tone wins over Formal Tone by a factor of two. We are very friendly. Our customers are very friendly. So we’re starting to tack some of those things, to be honest with you. It’s like, very new. I can tell you. People are enthusiastic about it. Let’s see how well it works and also in what uses we end up adopting it the most. Yeah.
George LaRocque: Well, this was a big launch. A number of capabilities, several capabilities that you’ve leveraged this technology for.
Nikos Moraitakis: But.
George LaRocque: As you look to the future, is there anything you can talk about that things that you’re thinking about, what it might mean for workable customers next?
Nikos Moraitakis: Okay, yeah. Let me start by giving you a little bit of background on non AI stuff. Okay. To put it in the context of the product and what it means for customers. Workable, from the very beginning, had a very simple philosophy about this, that hiring consists of a mixture of tasks. Some that you’d rather not do or some that are more suitable for a computer, and some that you very much want to do yourself. And when you want to do them, you have very good transparency. Our problem before starting workup was that as hiring managers, either we had to do a very unpleasant and bulky process or not get involved enough to weigh in properly on the decision. And in the beginning, we did it with better interface, with a nicer candid browser, giving more transparency, letting users collaborate more. That was ten years ago, and a few years ago, it was doing better, matching for you, automating things, doing yourself, scheduling, using templates, using triggers, using all sorts of things that start to take away a lot of the work that you don’t want to do and make you more productive. And AI comes in as, let’s say a finer layer on top of that because it allows us to go and do for you some things that previously it was very hard to automate, but now it is so beyond the magic of GPT as a general tool, it is mind boggling. But for the specific job, we don’t even want it to be too magical. We want it to be effective and reliable and just be able to automate some things that previously were unimaginable. So it fits very well with the philosophy of the company, fits very well with SMB. People often think that tools for enterprise should be more sophisticated. It’s the SMB people that want more automation. It’s the SMB people that don’t have hands right to tie it in with the rest. All of this for us, the copilot, as we call it, is the final word on here. We’re going to help you do the stuff where you need help with as to the future. As I said and as you alluded to, this does not antagonize users. It helps them, but the compilot helps you, how it shows you and suggests things to do and maybe participates in the doing. I think we’ll find that next year software like Workable will be able to take the copilot and say, here’s a few micro tasks, low level tasks that are not critical. Maybe you can decide that we don’t get enough candidates, so I think we should open up a premium job posting and make that decision. Or make the decision that, no, we shouldn’t hire an agency for this. I bet we can find or we should start this earlier because I know it’s going to do small things like that. There are other decisions like are we going to pay the location fees in the negotiation to get the developer? No, these are not going to be done by an AI. But you do have a hiring manager and a source persona already bring our business process, and we’re trying to take that source persona, which is high volume and needs reliability and consistency and is more suited for a computer with AI to do as much of it autonomously. I believe that in 25 companies like us may be able to announce, this thing is going to get you all the way to the first screening on its own. For some jobs, when you want under some conditions, it’s going to come and ask you for help sometimes maybe.
George LaRocque: Okay, yeah, that’s interesting. You’re right about the SMB. Talk about being resource constrained. These tools are critical in that environment especially, you talked about a few dozen to a few hundred employees. These are not big talent acquisition teams. It’s usually dedicated individuals to recruitment and so forth. And I love the idea of the system giving me some perspective about should we be sourcing, should we be advertising, should we be paying for premium? And the data is all there. And to your point on the SMB, getting it together, analyzing it, making that decision. A lot of times the time isn’t spent there because they don’t have it. So they repeat what they’ve done before. And I love that. Well, that’s a great view of the future. So I’ve got a question for you that I’m asking everybody and sort of back away from workable and this whole onslaught of AI. When I introduced the series, I said I was intrigued, I was excited, and then I was a little afraid. As customers or are looking at just AI in general, are there things that you would recommend that they think about or ask about as they’re considering implementing this into their workflow?
Nikos Moraitakis: It’s a really good question. Okay, I will try to answer it, but whatever I say may have changed in three weeks, but at some points, I think stan look, as far as I can tell from the knowledge we have, this new wave of LLM technology appears magical. Definitely has a ton of utility, but it’s a little bit like a very wild horse, if not a demon. Okay? Somebody needs to tame it for you to be useful and put a saddle on it so that as normal people can sit. So you need to make it reliable, you need to make it predictable, and to make it practical as well, you need to put in the data, the context and all that good stuff. So my experience has been that after I was amazed by it, then I said, how can I make it, know my stuff, do my work? So whoever as a software vendor or whatever other role, whoever can cast that context, give that context, and pass and apply the technology onto it has value to offer is going to be important. Not the most important, but definitely important in the process. And if a tool isn’t doing that for you, doing it yourself is not an option. Really. Like the engineering that goes behind you. See this job description? This is getting fed examples that are created by other AIS that are look at it’s given constraints. Don’t do this, don’t do that. There is a special LLM that is just monitoring the outputs to see if they’re saying anything crazy. So somebody needs to provide that context for you. And that’s going to be the initial value of vendors like us. Where are we going to get them? Where are we going to buy them for my company? Expect my existing vendors to do it. I expect Google is going to add it to our Gmail so that it’s ironclad and it’s also in a safe context, and it also inherits the privacy, and also so that it is something that reinforces. Right now there’s a lot of GPT wrapper and a lot of, oh, look, I can eliminate an entire category of things with a clever thing I did with CPT, but when you go into depth, these are not going to be universally applicable. So my instinct to viewers is like, what’s the technology? It’s an incredible productivity boost. You definitely want it, but you also want to find, ideally, your existing vendor to come and bring it to your hands where you need it to tame it for you and to put it inside the cage for you and to be watching over it.
George LaRocque: That’s good answer. I think there are a lot of implications here with ethics and culture and thinking about your data.
Nikos Moraitakis: I love.
George LaRocque: Getting a saddle on this horse. That’s a great analogy. Nicos, this was great. I really appreciate your taking the time to walk us through what you’ve announced and what you’re thinking about with AI and your success so far. I’m going to take you up on that offer to come back in a few months and check in. But thank you so much for being here.
Nikos Moraitakis: Thank you, too. The pleasure has been mine, and all the best for the podcast.
George LaRocque: Excellent. And thanks to everybody watching or listening. And until next time.