Processing 20 Million Minutes of Audio and Video Speech-to-Text With Sebastien Pare

SlatorPod #185 -Processing 20M of Minutes Speech-to-Text With Sebastien Pare

Sebastien Pare, CEO of VIQ Solutions, joins SlatorPod to discuss the evolution of transcription for multi-speaker content in challenging legal and regulatory settings.

Sebastien explained that VIQ initially started in Canada and expanded to other Commonwealth countries, focusing on courts thanks to their need for advanced audio and video solutions.

The conversation delved into the challenges of working with multi-speaker content and Sebastien highlighted the importance of AI in enhancing transcription accuracy, especially in industries where the quality of the source material varies widely.

Sebastien addressed linguistic challenges in the transcription industry, including dealing with varying speech quality, jargon, local dialects, and slang. He highlighted VIQ’s flexibility in working seamlessly with third-party technologies and its commitment to offering various pricing models.

Subscribe on YoutubeApple PodcastsSpotifyGoogle Podcasts, and elsewhere

Sebastien also discusses the challenges of operating a public listing (the company on September 30, 2023, announced it will stop its dual listing on Nasdaq but remain listed in Toronto). He also touched upon the impact of the AI hype, buyer expectations, and the macroeconomic environment on VIQ’s operations.

The discussion shifted to the shortage of court reporters and transcriptionists, where Sebastien explained how VIQ is attracting younger talent from diverse backgrounds and leveraging AI to increase productivity.

Sebastien shared some developments in VIQ’s roadmap as they plan to enhance their summarization capabilities, and further localize their platform for different regions and languages.

Transcript

Florian: Sebastien is the President and CEO of VIQ Solutions. VIQ Solutions is a provider of AI-driven digital voice and video capture tech and transcription services. Hi Sebastien, thanks so much for joining today. So all right, tell us a bit more about your background, journey to becoming CEO of VIQ and just maybe what VIQ does.

Sebastien: It’s been a passion of mine to basically build organizations to solve real kind of real life problems and it’s always been around workflow and it’s always been around productivity enhancements of the human brains to try to go a little bit further in terms of productivity. So I’ve built other businesses, I’ve helped develop businesses. I particularly came from the kind of case management software. That’s where I spent my first 12 years, so really in depth knowledge and time spent with customers, particularly business-to-business and business-to-consumers, on helping them to build basically process improvement programs. And you’re going to realize really quickly that the workflow is at the heart of any productivity gains that we’ve ever witnessed in any kind of organizations moving forward. So that’s really where it started for me and then obviously came across VIQ to the work that I’ve done with government agencies around the world, but also in courts, because courts is really archaic in terms of technologies and everybody saw it coming, particularly when it comes to the audio and video, as well as the production of the incredible amount of documentations that need to be produced as evidence in order for this workflows to work. So came to that angle and then really, honestly, at the time was connecting all the dots with VIQ was the biggest opportunity of a career in an archaic industry. That the last time they saw any kind of modern ways of doing things was when the keyboard and the shortcuts were created to allow the editors and the transcriptionists to do their work better. So it’s been a journey, it’s been a stage, and it’s always been centered around pushing the productivity of the human being and the experts that are really involved, servicing those customers and clients and how to basically solve their problems and trying to make them a little bit more productive as we’re going forward.

Florian: In terms of the courts, I mean, since you’re in Canada, was it primarily initially in Canada, but then very quickly expanded into the US? Or was the US and Canada kind of at the same time, were these the two focus markets and anything in Europe? Because, of course, the court system is very different across these jurisdictions.

Sebastien: They’re very different. Obviously it’s been part of the Commonwealth kind of countries to start with, but it started in Canada, particularly with the courts, and it really originated from the audio and video recording that we were just talking about before the show and the ability for us to be centered and focused on multi speakers. I think that’s really, really critical. VIQ expertise always been pioneer around the multi speaker. It started with audio and video recordings in the courts, where you could have up to 10, 20, 25 people being part of a process for multiple days, all the way down to one or two individuals involved in a case, foreign insurance companies who are trying to settle claims. So the multi speaker aspect and the expertise of VIQ has always been the core of the business and then we evolved that in around 2018 towards… Well, once we’ve mastered the multi speaker recording up front, then the ability to start building the right type of technology, i.e. the platform in order for allowing our customers and clients to start producing all the transcriptions because up to that point, it’s always been done manually with a footset, a headset and a foot pedal and two screens and it’s all been manual for decades. And so the opportunity was there and I think the market was evolving in that direction.

Florian: Very interesting. I got to admit I have very little knowledge around that particular process. When you say the foot pedal, like, what would that be used for?

Sebastien: It’s the ability. Whether or not you’re in Canada, whether or not you’re in the UK, because VIQ is a global entity. We started small, but we’re now basically got more than 50% of our revenue in Australia. We’re growing quite steadily across Africa. We also have a growing presence in Europe, particularly in the UK, where we’ve got a good base, and obviously United States and in Canada as well. So there’s differences when it comes down to audio and video, but in terms of production of the documentation, you can appreciate that the archaic way is you upload a file, you basically have multiple screens, and then you have to take the time, one speaker at a time to start producing the transcript. And then you basically use your foot pedal in a way to stop every couple of seconds so you can write things down, you could go back, verify, you can appreciate this manual process that has gone on across the industry globally. And so everybody knew at one point you combined that with a shortage of stenographers, the shortage of court reporters, which we’re going to cover to this podcast, the timing was absolutely ripe for some kind of a technology modernization and AI is really at the center of all of that with the evolution of the language models and the speech-to-text engines in the last couple of years.

Florian: 100%, so that was a big kind of big moment, I guess in the last 12 months. But since you started as a CEO of the company in 2015, can you just take us through some other milestones, memorable moments, kind of new tech, maybe business wins, challenges that emerged, just share some of those episodes.

Sebastien: I think from 2015 when I came in, company was already public, so it is what it is, but there’s so much that has changed since 2015. So I think for me one of the key milestones is we created the platform that we set to do and we use our bays in audio and video recording for the courts and we diversify into law enforcement, we diversify into insurance, we diversify into media. So we took our expertise in multi speaker courts, also creation of document, to expand globally, knowing that the real needs were no longer on the recording side and the real needs were really shifting really quickly on addressing this massive problem really to this manual way of doing documentation, particularly as the markets were just expanding drastically. So we created the platform we set up to do. We know it works. It’s been proven now and then it does improve the productivity. I mean, in some cases, Florian, we’ve noticed a 50% productivity gains and obviously courts is really one of the strongest kind of market for adoption because there’s good quality of audio and video coming in, so a tremendous process that we’ve gone through. But we’re now at that stage where we know it’s been working and you can see the evolution of the company in terms of productivity gains and how it impacts… margin. So we also completed the acquisitions that we set up to do. We’ve entered the segments that we set up we’re going to do, and we did it well. We’ve integrated those companies and now we basically also were able to train, we’re talking about thousands of traditional manual transcribers. They’re now professional editors on a cloud-based workflow platform that is driven by AI and they’re collaborating with the machine learning in the creation of a first draft on behalf of our client. So all that work that I just described took about five years to go through because first you need to build the tech. First of all you need to have the expertise, build the tech, build the platform. Then you need to build your cloud infrastructures and then you became the migration of all your customers into the platform. That’s kind of stage one. Stage two was the human element part of it, and the ability to bring people along that AI is not here to replace the editors. As a matter of fact, it’s here to protect the editors, giving them a future in an industry that is undergoing a tremendous amount of changes. So we’ve been really proud of that, that we’ve been saying all along that if we’re going to be there early, we’re going to lead this transformation. But it’s really about bringing our people and our clients along in the idea that really at this point, that there’s technologies now available that could actually improve tremendously how things are being done. So really proud of that, and obviously, throughout this whole process, there’s been a lot of noise. I mean, we’re the only kind of publicly listed company in our space, so I compete with others that are privately held. We’re public, which means everything is available publicly and things like that. So we’ve gone through the public ups and down, we’ve gone through COVID, we’ve gone through the great resignations which impacted this industry massively, as you know. But we’ve managed to really deliver on what we said we’re going to do when it comes to the technology, the platform, the improvement in the productivity, and basically the ability to bring people along and cross train to become professional editors. So I think that was really important and just leave you with that specific statistics. But for me personally, as of December 31st of last year, we basically process almost 20 million minutes of specialized audio and video content and we produce basically almost 6 million pages through the platform, so you can appreciate the complexity of those pages. They represent court cases, court cases with 10, 15 people that have gone on for 10, 15 days, highly regulated content. So it’s something to become quite proud when it comes to scalability because now we’ve gone global and I think at this point we’re really excited. We’re in the process of finalizing the migrations of all our core customers in Australia. So we’re about to begin the next step in terms of showing how that platform is transforming the industry.

SlatorPod – News, Analysis, Guests

The weekly language industry podcast. On Youtube, Apple Podcasts, Spotify, Google Podcasts, and all other major platforms.

SlatorPod – News, Analysis, Guests

Florian: All right, so you were talking about the editors, so those are, I’m sure, experts that are editing. So can you just tell me a bit more about the profile of these editors and how they kind of work on your platform?

Sebastien: Yeah, so clearly we’re focused on verticals or industries that required a really high level of accuracy and usability. So there’s reasons we’ve entered and we are dominant in courts, insurance, legal, media and government and criminal justice, because they all have one thing in common, that the content that we need to produce is multi speaker. Accuracy is paramount because if you look at the outcome of what we produce, it feeds into a broader case management workflow where the outcome will be dictated and driven by the quality of the evidence material that gets submitted, so accuracy is really critical. So if we look at the editors involved in the process, as we mentioned earlier before, everything was done manually in the past. The best kind of technology was… it comes up with a template and then it’s all done manually and it takes a lot of time and it’s a lot of effort. So if we look back at who used to be an editor and the kind of profile versus today, things have changed. There’s a shortage of this kind of expertise globally. We’re not alone. I mentioned earlier the great resignation that has caused a lot of trouble in the industry as well. So really, if you look at the profile, the profile has gone from someone that was doing everything manually from a blank page all the way now to they’re basically working in parallel in collaboration with machine learning and now a first draft comes up, so we’ve built up a whole lot. We’re up to 42 specific productivity tools built into the workflow platform that allows our editors to actually perform at the highest possible level. So if you look at who they are, there’s more and more. They’re part of the gig economy, so a lot of them are contractors, a lot of them are now coming in from the film industry, directors, basically clerical staff in process to become lawyers with law firms, all kinds of people that are actually now in the process. So the profile has evolved tremendously. A tremendous amount of the command language, whether or not it’s English, French, Spanish or some of the other languages that we get involved with. So extremely high level of the command of the language, combined with the ability to really work at a high level in terms of productivity, with the machine learning, editing, and then you’ve got a whole list of requirements that is unique for each customers. So the editors that we’re working with, they get trained. I mentioned earlier, we’ve trained thousands of them over the last couple of years as we transcribe the work that we do and the profile has evolved and it’s getting a little bit younger. So everybody is thinking about the court reporter. You mentioned one of your questions, there’s a shortage of court reporters, what’s going on there? How is that impacting? And obviously it’s impacting everybody. But what people don’t talk much about is a whole new generation of editors where it’s no longer about AI is a threat. They actually want to put on their resume the ability to collaborate with AI. They want to be part of that process and the independent workers, the work flexibility, the hours and all of that, so that’s all part of that process, but we go global. We find those skill sets globally, they get certified, they get trained, and then they get into the platform. There’s also because of the nature of the work that we do, Florian, there’s a lot of security clearance so that’s another key differentiator compared to any other verticals. We’re not in a vertical where just everybody could actually see and watch that content. Whether or not it’s criminal investigation, whether or not it’s civil proceedings in a courts, it’s highly confidential. So we have a whole process internally to qualify and make sure that our editors have all the security clearances in each of the region where we operate in order to have access to this kind of content and being part of that process.

Florian: Yeah, that sounds like a pretty big challenge in terms of logistics. Now for kind of the linguistic challenges, what are some of the major areas there to focus on in your areas? Again, law firms, criminal justice insurance when it comes to speech recording, quality jargon and languages. I guess most of it is in English if you’re in US and Canada, but talk to us a bit more about the linguistic challenges in your business.

Sebastien: The challenges are variables across all the industries. If we start from the top, I could tell you that the quality is ranging from very very good in the context of courts and legal proceedings and it’s really, really low and in some cases pretty bad when it comes to investigation, someone who potentially is being interviewed on the street or in a precinct or in a police office station for an interview. There could be drugs, there could be alcohol involved, so the language gets a little bit tougher and tougher. So from industry verticals’ perspective, I will say courts and criminal and legal are really at the top because most of the organization have invested in high quality recording solutions. So the quality of the audio and the video that is coming our way is really good, obviously. And if you tie that back to the productivity enhancements that I mentioned earlier, that’s why we’re seeing a tremendous amount of uplift in the productivity because now our AI and our workflow is able to produce a fairly accurate and highly usable first draft that is making a tremendous impact. And at the lower end of that you’ve got kind of law enforcement/criminal justice where some of that is pretty tough. So it could be there’s multiple speakers involved and then they are going to say yes or no and then they come back an hour later, they say yes or no. So we need to be able to distinguish, so that’s a challenge. If they don’t talk much, the ability to diarize and recognize that voice on a 15 hours proceeding becomes a little bit tricky. There’s words where depending on the context, basically it might mean something different in industry and it might mean something else in a different industry as you’re looking for it. So we do a lot of normalizations and that’s all part of our AI language models that we get involved with and all of that. Then you’re getting into the slang, you’re getting into the jargon, you’re getting into the local dialects that are really tremendous pressure points and then we’ve built different tools on the AI to also accommodate that. But really it comes down to, at the end of the day, that every industry, every region will have its own challenges. But it’s a matter again back to the workflow to build post-processing, to build knowledge repositories from the edits by the human brain that is working on the platform in order to basically build models that allow us to really normalize all of that, diarize all of that and the ability to really bring forward an audio that is as normalized and as clean as it can be. At that point, the other part of the workflow will pick it up and then we actually start producing the actual documentation. So we’ve got a tremendous amount of complexities, that’s why it’s not an overnight thing. That’s why if you look at Google, Microsoft or anybody else, they’ll never get to a point where they get customized for the verticals that we serve, because it’s pretty deep in terms of what’s required to really improve the quality of the audio and the video up front in order to give you a chance to produce a fairly accurate first draft.

Florian: Again, what you described as an incredibly complex kind of expert-led model, tech meet service, but when something like Whisper came out, must have been like nine, I don’t know, 12 months ago, and this whole AI hype starts gathering pace. I mean, how did you think about it at the time, how are you thinking about it now? And what do you think kind of are the implications for the wider industry there?

Sebastien: I think we called it early, early. That’s why we focused on the workflow and not speech-to-text engine. It was a matter of time and kudos to obviously OpenAI, but Whisper was truly the first open source that really from a commercial grade perspective has allowed us to leapfrog a whole lot, so we’ve used them all. We’re agnostic to all speech-to-text engines. We’ve gone through, I think six or seven so far. We have a methodology internally where we monitor, we score for each of the use cases by region, by vertical, and then we basically reroute the file based on that scoring to what kind of speech-to-text engine we’re going to be using. But remember, speech-to-text engine, really, it comes down to word, I call it word accuracy error. It’s been commoditized. I mean, it’s just phenomenal, word for word, what ASR engines have been able to achieve. But we’ve said it early, that day is going to come and that’s what for us, it was a matter to become agnostic to all of that. And sure enough, when Whisper came out late last year, we were one of the first company commercially to file patents related to Whisper. So now we’re deep into the Whisper world. We have an ecosystem with them that is pretty deep in terms of we built all our language models now around that and obviously our overall production cost obviously has come down because it’s a game changer in terms of what Whisper allows us to do and now we have complete command of our own models off Whisper. So we’re still leveraging the large data model and the large language model of the Whisper of the world, so as many others. But we’re kind of focusing right now on Whisper, but we’ve built our layer of models that is unique to us, which is really the outcome of all the edits and all the work that our professional editors are actually doing on the platform 24/7. We’re repurposing all those edits and correction and we built learnings into the model and that’s why our FirstDraft product has gained so much traction within the customer base. Everybody could run speech-to-text, everybody could do a streaming technology. The word accuracy error, whether or not you’re on a commercial ASR engine or on an open source like Whisper, is just phenomenal. It’s no longer where the issue is. The issue is really, what do you do next? And the ability to deal with complex formatting, multi speaker diarization and the ability to deliver something that is highly usable and highly accurate and that’s what we’ve been focusing on.

SlatorCon Silicon Valley 2024 | $ 1,340

SlatorCon Silicon Valley 2024 | $ 1,340

A rich online conference which brings together our research and network of industry leaders.

Buy Tickets

Register Now

Florian: Yeah, you mentioned FirstDraft, which is one of the products that I saw on the website as well. You have a number of other products like CapturePro MobileMic, NetScribe, aiAssist, et cetera. Is FirstDraft where your focus lies or is it kind of a complementary product range? Tell us a bit more about that.

Sebastien: Yeah, so we’ve got different kind of products, but it’s kind of mimicking the workflow associated with whether or not you’re on an insurance doing a call because it’s a settlement and it’s a claims or it’s a disputed record, whether or not it’s courts, whether or not it’s a police officer that will stop and need to do an interview or it’s a media or earning calls, we kind of mimic that process. So we have products whether or not it’s CapturePro, CapturePro Mobile, which is the application for on the mobile side of it, where we’re basically built-in expertise from multi speakers, so we’ve got technology and products related to that. Then it basically goes into the core of the platform, which is called NetScribe and NetScribe is the connecting tissue between the audio and the video on the front versus the ability to produce documentations at the end, so NetScribe is there. aiAssist is the AI engine, kind of the black box that sits underneath NetScribe where all the magic happens. That’s where all the investment has gone to. FirstDraft is a core product that comes out of NetScribe, which is kind of the first draft in terms of the content, but that first draft gets consumed. We’ve got our own editors. We also have a number of growing lists of editors that are not part of the VIQ ecosystem. They’re using NetScribe for their own purposes, their own transcription needs, so we’ve got all of that. On the case of media, we have an AI technology that is self serve, that allow the journalists, once the content is produced, to be able to query, search and generate their own content. So that was a very specific targeted AI investment that we’ve made and I’m glad we did in terms of the media piece. And AccessPoint is the last product, which is really a portal that allows our customers to manage their content and then whether or not it’s the recording, the documentation that they produce and all of that, so it’s basically a suite of products. They’re all kind of connected properly and then for the other customers where they might be using someone else’s products. Whether or not there’s a lot of good products out there for other court technology for recording, it could be insurance recording, so we’re not alone in that particular market. And what we’ve done is we’re also agnostic, we work equally well with anybody else’s technology on how we’re capturing the audio and video. We’ve built a set of APIs that allows that content from a third party to be able to flow very effectively into the platform. So that’s kind of the platform as a service that we’ve built over the last couple of years.

Florian: It’s all kind of enterprise, I mean enterprise-grade, of course, but also on the pricing side, you don’t have kind of a light version that people can subscribe to in like a monthly SaaS pricing way?

Sebastien: Yeah, we do, so it’s the platform. The core of our business is with large government agencies, large law enforcement agencies, large Fortune 500 media companies. So that’s on the heavy side of it, where the bulk of our revenue is. But we have a growing segment of our revenue that does come from much lighter access to some kind of SaaS. So we have built in a number of different price points for different type of users and customers to start onboarding into the platform as well.

Florian: Got it, so you’re publicly listed and so you have the pleasure of talking to analysts every quarter, I believe. And so we looked up the most recent conversation you have with some analysts and in your second quarter financial results, you mentioned that you’re seeing record new organic bookings, about 9 million. But also you said that new contracts are taking longer to ramp up due to, you said, project and IT capacity within all the big enterprise clients you’re seeing. Now, in terms of kind of dynamics here, where do you see this kind of AI hype? And I think that might kind of lead to buyer anxiety, maybe over committing to new language related tech and then kind of on the one hand, on the other hand you have maybe a general kind of macro slowdown in the economy. I’m not sure if this is even relevant for your business because well, I guess the courts work no matter how well the economy works. But how do you kind of weigh these factors kind of AI hype, maybe some buyer anxiety and then kind of the general macro-environment?

Sebastien: That is definitely a challenging one. So you brought up the public company, you brought up what’s going on in terms of the customer base. So maybe I could just try to break it down. So from a public perspective, public company perspective, that’s why I’m smiling. It’s a challenging, challenging journey. The company was public when I came in so that was not my choice. And I could tell you at this point it was one thing and there was a lot of positives when the markets were up. But obviously when there’s a market correction as severe as what we’ve seen post-COVID, I don’t think anybody saw it coming. But it’s been absolutely brutal to get to that correction, so it is what it is. We don’t control the market. We’re trying to focus on our own… On what we can control, but there’s a couple of things coming up on that. One is you’re going to hear more CEOs like myself that the cost of being public is outrageously high and while everybody’s focusing on profitability, good shareholders will demand all of that and we respect that. But the threshold for a public company to be profitable is insane. Just to give you a sense, we’re approaching eight and a half, almost nine per cent of our total cost is related to being public. And auditing is one thing, regulatory filings and also the disclosure that you just talked about. And as part of your process, I’m sure compared to any of your other interviews, as a public company you have access to everything and then every three months we basically have to stand up out there and tell them how we reinvented ourselves in the last three months and how much growth is coming in. So for me as a CEO, it’s always been the challenge. It’s a balancing act between short term results and short term expectations versus a bit longer term and what’s right for the business. And everybody will have a different opinion in terms of that. I’ve got people that are in the company as owner for less than three days. I got some people that are in for a year. I’ve got some people that are in for 10 years and they all have different views. And then you’ve got the analysts that are coming in that are trying to assess you every three months, so it’s a challenging aspect. I think we’ve done pretty well to deal with the disclosure aspect of it, but it’s a highly multilayer challenge and for me, overall, it’s the cost. Number two, it’s the time horizon because now you’ve got a large family of owners and every one of them seems to have some kind of a different perspective. So our job is trying to kind of stick in the middle. So to your second part of the question, that was a really good one as well, is what’s the hype and what’s going on. So maybe one way to explain it, I was thinking about it is, you know a company called Nuance? They were acquired by Microsoft last year and all of that. I just wanted to kind of bring you back in your audience. 2017 and then basically Dragon basically early stage speech-to-text engine came with a bit of a lighter platform on the back for medical transcription and then they had this major, major malware event that happened in 2017. The whole world was aware of it and then obviously everybody was wondering what’s going to happen next. Well, in a way, what Nuance experienced in 2017 with the malware event is basically what the shortage of court reporters is driving that level of disruption in the industry that we’re in today. There’s nobody with a bit of a rationale mind that could stand up in 2023 and basically said that this multi speaker, eventually content production, documentation is going to remain the same that it was ten years ago. I think we all came to terms with that. There’s a few companies like us who obviously took the lead early on, but that malware that kind of basically created a crisis in the healthcare industry back in 2017, in a way, the court reporting shortage combined with the AI that is now going on has now created a perfect storm where now you’ve got customers that are now saying we’re going to accelerate our migration to the cloud. We’re going to basically change our workflow because what used to work 30 years ago, 20 years ago, is not going to work moving forward. We’re not able to recruit enough people with the skill set in order to do the work. So everyone from us down to our customer, it’s really about trying to improve and increase the productivity of a shrinking expert global workforce in that space and I think that’s really what’s going on. And I was thinking about the best analogy to bring it up to you, but I think that event in the healthcare industry in 2017 is what really in a way, is the equivalent of the court reporter shortage, expert shortage and the global workforce shortage that we’ve experienced in the last two years. You combine that with the massive evolution of AI in the last two or three years, and it’s getting better and better and better, that I don’t think any clients at this point could stand up and said that we’re not going to be modernizing the way we’re doing it because they won’t be able to service their customers moving forward. The backlog is getting bigger and bigger and I don’t think they could easily turn around what used to be a five-day turnaround will now become a ten-day turnaround. Clients are expecting an almost real-time turnaround. They’re expecting a one-day turnaround.

LocJobs.com I Recruit Talent. Find Jobs

LocJobs is the new language industry talent hub, where candidates connect to new opportunities and employers find the most qualified professionals in the translation and localization industry.

LocJobs.com I Recruit Talent. Find Jobs

Florian: What triggered the court reporter shortage beyond just the COVID I guess, general, yeah, you said great resignation? Was there anything specific about court reporting that triggered this?

Sebastien: I just think it’s the expertise. It’s hard work and if you look at the skill set to do all of this manually, you need to have an extremely high level command of the language that you’re operating. You need to be super, super sharp when it comes to manual efficiencies from the brink. I mean, we’ve done the study in terms of how to connect the audio and down to the keyboards and all of that. So that generation kind of doesn’t exist anymore. So if you’re looking back, nobody’s going to school to become a transcriptionist or court reporters anymore. So what we’re seeing is the evolution and what we’ve seen with other industry goes through a modernization on how that process is being handled. And I remember doing a couple of interviews a couple of years ago and we were accused of replacing jobs and taking jobs away. All of this has gone, it’s now about really providing a future and sure enough, we are now recruiting what used to be an average of between maybe 40 and 55, 60 years old. We’re now basically recruiting people in their 20s, in their 30s and they come from a different walks of life. I mentioned earlier. Legal industry, the film industry, the media industry, writers, all kinds of individuals that are basically using that process as part of the gig economy to not only have a future, but also to make good wages. And potentially make the same amount of money that they were making before, but with less time and then much higher level of productivity, as opposed to sit in an office and having to start from a blank page every time. So there’s been a real change in the way this industry has evolved and I think that’s really what contributed to what’s going on today.

Florian: We need new names also. I don’t know, like AI-enabled legal expert or something like that. I mean this is a terrible suggestion, but something that makes it cool again, right? Also, obviously Slator is very deep on the localization and translation side of things. These old names for jobs probably just won’t do justice to what these people are going to be doing in the future. Actually, most of them already doing now, actually.

Sebastien: It feels like you were a little bit of fly on the wall. We’re just going through a process where if you’re looking at, to your point, about how those experts were classified. The existing classification, Florian, actually came from 20, 25 years ago and you still have word like Typist, you still have word like QA Reviewer, like it’s all geared towards that old fashioned archaic manual process that was going on. We’re now in the process of revamping all of that, but it’s in the context that nobody is starting from the blank page anymore. And now half of what they’re doing is a tremendous amount of interaction with the platform and then behind the scene there’s a tremendous amount of learnings that are going on which gets repurposed. So now there’s a new level of classification, there’s a new level of productivity. How productivity is measured is very different today than what it was a decade ago. So to your point, I don’t think we’re the only industry, but we’re clearly on the front line of how AI is having a meaningful impact. It’s not a hype, it’s actually real and it’s actually gone on now for a couple of years and then we’ll come out of it and in a way AI will be normalized. And that’s what we’ve seen from my previous days in case management with other companies. In those days it used to be called expert systems, it used to be called business rules oriented, trigger points. We were doing everything we could to augment the productivity of the human brains. But today, obviously we’ve moved forward into something else but the learnings behind the scene. I mean, I mentioned 20 million minutes. We’re going to be crossing the 30 million minutes this year on the platform and then that represents millions and millions of edits that now the human brain is involved with. And that’s why the language models that are being built that are unique to the IQ by region and by vertical is really starting to come up as the differentiator because every one of our clients has been running speech-to-text on their own. A lot of our clients are easily capable to run streaming and yet they’re not able to automate and they’re not able to create a first draft that is accurate and usable enough. And I think that’s what really comes into the complexity of what we provide.

Florian: Now to close anything you can share about kind of exciting developments, innovations that VIQ has on the roadmap going to 2024 that you can announce? Understand public listed company, there’s only so much you can tell us ahead of time.

Sebastien: For us, really, there’s been a couple of things, so we’re extremely excited about it. Our roadmap, I’m a technologist at heart and as far as I’m concerned the roadmap is that vision and that long term ability for the company to continue to perform. So we’ve added video, which is a really big deal. If you think about some of what we’re just described in the last couple of minutes, we’re talking about massive files on the video side. And the video is really required because when you come down to the language, the audio versus the video, the ability to reconcile both has a tremendous impact on the accuracy of what our editors will hear and thus the ability for machine learning with the editors to get it right. So video was added last year, a tremendous amount of improvement in terms of what we’re doing. Obviously the body cams certainly pushed in that direction, but now everything is being digitized and everything is being recorded. So audio and video is now equally dealt with for us. I mentioned we’re up to 42 very specific productivity tools built into the platform that allow editors to perform at the highest level. We’re going to double that number in the next 12 months in terms of next wave of very unique tools based on the platform to allow that. But one of the things that I wanted to talk to you about because of Slator, we also have tremendous, tremendous demand. Already today a significant portion of our revenue goes to what we call foreign languages and multi-languages. So now we’re talking about Whisper. We’re now in the process of basically adding the ability to have multiple languages as an offering to our customers and the uptake has been phenomenal so far. Obviously it’s English to other language, but it’s also other language to other languages as well. I mentioned Africa. We’ve got a tremendous, tremendous track record to meet the demand and that comes with different languages. We’re involved in Canada, where you got French and English. We’re involved in North America, obviously with Spanish, but our presence in Europe has been growing, so now we’re basically diversifying in terms of multiple languages. So multiple languages, stay tuned. But it’s becoming a very critical aspect of where we’re going next and it’s already happening at the platform level. So now we’re getting into that process big time this year. The other piece that will not be a surprise to you, coming in from Slator, but the summarization is becoming really, really critical. We now reach a stage, we have enough scalability, we have enough accuracy in our content, we have massive data lakes that we’ve created behind the scene, by region, by vertical, that now we’ve got the ability to turn on summarization and obviously there’s a number of really sophisticated tools that we brought in. So now we’re offering different level of summarization. Some judges, I could tell you in the UK have different levels of requirement for summarization that is different from a police officer that is doing a criminal investigation. So multi-languages is a big deal for us this year. Summarization and then localization. What I mean by that is the platform itself now has been localized in other languages. So now we have reached a point where we’ve got customers in other regions with different languages that want to interact with the platform in a different language. So the platform itself is also now being localized for different regions with different languages. So those are the main components, but it all ties back. We’ve got the data in the first place. We’ve got the audio, we got the video, and we’re spending a tremendous of effort trying to produce the most accurate documentation we can get. Out of that process, we do a whole lot that is aiming towards our customers now, are asking us, as you’re going through that process in a secure way with security clear people involved in the workflow, could you also tackle some of my multiple languages needs? Could you also make some of the summarization available to some of our stakeholders so we could actually improve the overall case management software that they’re using at the other end? So those are the main buckets of what we’re working on for this year.