Hello and welcome to the Translation Company Talk, a weekly podcast show focusing on translation services on the language industry.
The translation company talk covers topics of interest for professionals engaged in the business of translation localization, transcription interpreting, and language technologies.
The translation Company Talk is sponsored by Hybrid Lynx.
Your host is Sultan Ghaznawi with today’s episode.
Hello and welcome to this episode of the Translation Company Talk Podcast, we are back after a short break as I was very busy in September. September and October are generally busy months as people return to work from holidays.
Conferences are in full swing and work piles up quickly.
In any case, we are back with another episode, and today I’m excited that we will be covering the subject of embracing.
Automation in the language translation process.
I’m sure you all know that AI has been around in the form of machine translation in this industry for what it feels like a couple of decades already, we will be exploring automation beyond that, in our conversation today.
My guest today is Matija Kovač.
He has successfully founded two companies, a language school which became to be the biggest language school in Slovenia and a translation platform.
His first love is languages. He has studied Chinese language and spent some of his time studying in China. His other love is development. He’s been developing since high school and is the development father of Taia.
Matija and Marko founded Taia to enable a more secure and time efficient translation process, seeing that sending files over email can be a security threat. They decided to create an app with a huge emphasis on keeping the files secure.
Since its first launch, the Taia app is consistently developing, creating new features to enable companies and an easy ordering process, while being able to oversee all projects, the translation process is supported by the latest translation technology, which enables getting translations up to three times faster than usual in a cost-effective manner.
Welcome to the Translation Company Talk, Matija.
How are you today?
Hi Sultan, thanks for having me.
Please tell us about yourself and what you do?
So hi, my name is Matija. I come from Europe from a small country called Slovenia, and I’m by education language professional, but by profession I am a developer. I am head of development at Taia Translations, the startup that I co-founded.
And we are focusing on building a platform for automation in the localization industry. So we try to help clients who need translation services and get them as fast and as efficiently as possible by leveraging high level of automation with a human in the loop process.
Very, very exciting. I would love to hear a lot more about that. How long have you been in this industry, Matija?
So, I started working on my first startup in 2014, and it was a language school and a few years after we saw that a lot of our clients were looking to us if we can handle their translation services as well. So thinking about 2016 something like that, we started working more seriously.
And providing translation services and then in 2018, we realized that there’s no way to start a business or there’s no need for another LSP on the market. But there’s a huge demand for there’s a huge gap between what the technology can allow and can be done with, and what the market is actually offering.
So, that’s how we started working on what is now known as Taia.
Please share a few words if you will about your journey in the localization space. What were some of the most significant observations that impressed you speaking from a technology perspective?
Of course. Right so, as I mentioned already, I’m a nerd. I love computers and everything related to decoding, and I’ve been working on this for a while, so it was very shocking to me when I learned back in the day that a lot of the translation industry still relies on sending content around via email.
Sales and preparing quotes for clients in so like software that requires a lot of manual work. Sometimes companies even still use Excel to build a quote for each client separately every time. And this is like the amount of wasted human time and potential that I saw in the industry was what drove me to start working on this platform, so with my initial idea that we should automate as much of this as possible to allow for people to have a better lifestyle, so they can work on things that are more interesting and not do repetitive monkey tasks every day.
Let’s talk about the topic of our conversation, you mentioned automation. It has made significant progress and translation and localization. Almost all of us have embraced it and executing translation work
Mission work tell me how are we using it in processes around translation and the language services.
Right, so yeah, when we started there was like new machine translation was something that was almost brand new. If you remember it was, I think it was 2016 when first actual neural machine translation engines came out on the market and adoption was very slow in the early stages.
There was a lot of machine translation already in the past, but mostly it was based on older technologies like statistical models. And even in the recent years, finally companies are catching up and starting to use this.
This is just one part of the entire translation process, so if you imagine how a client would order their translation services within LSP, a lot of this is what the translator are doing and this is now finally being, you know, speed up with all these systems from neutral machine translation, so translators can work faster.
But then there’s this whole part all around this that’s not done by translators or other vendors, but it’s actually done by project managers and administration and vendor managers and so on.
And this is something that, in my opinion, also has a huge potential for automation. We can make these things much faster.
So, if you were to identify at a high level what areas are mature to be automated within the language services delivery space, what would they be?
Definitely, I think processing of files is something that we’re doing very well in our end. I can’t speak for other competitors in the market, but I know this is something that our clients really like that when they drag and drop a file into the app immediately in seconds they get a fully analyzed document of a full analyzed report, so they know exactly how much translation memory is being used, how much are they going to need to pay for the project, and even when it’s going to be delivered. And this is one part that’s not automated with all the providers or most of them, and something that works well for us.
This is something that our clients like because it’s very transparent. They just drop the files, and they immediately know how much it’s going to cost them and when they can expect it. But then there are other areas as well, like how you manage the project, how you manage your vendors, how you manage your administration. We have a lot of already in place, but still there’s a long road ahead of us of what else we can automate.
So, please explain how do you see applying automation and the project management models that most translation companies use today?
Well, this is so… I don’t want to explain it too much, maybe I would give away some company secrets.
But one thing we’re looking at is predictions on which translators are most likely to accept a job and are going to also perform well on this job. And this is the part where we’re looking into how to automate. We don’t have this fully solved yet, but it’s definitely an interesting area for us, because you lose even less time by allocating a job to the right translator, if you are able to predict whether or not they will be available or not for this task.
But then there’s other stuff also in the vendor management process, because when you start onboarding thousands upon thousands of vendors, there are certain tasks that could be automated and speed up in order to make sure that you are working with the right people at the right time, right?
So, when it comes to production models, uh, there are so many things that can be automated. It’s not just simply the task of, for example, the handing off and handing back in of the work. Can you elaborate a little bit on what specific activities have you been managed to successfully automate?
Oh, that is a difficult question, because there’s so many little points there, so many different things we do, and there’s so much more we have planned still, but how to go about this?
Well, definitely one thing that saves us a bunch of time, and our clients really appreciate, is the full or not needed updating of translation memories depending on what’s going on with the project, so our PM doesn’t have to do as much work on gathering the right teams and cleaning up those teams and pushing them back into the system.
This part is mostly, if not entirely, automated in our case. So as long as we handle the process inside our platform, which means that everything is translated, revised and proofread inside our ecosystem that the teams are going to follow this entirely automatically, and you can see it all in different stages. So, there’s quite a big advantage here.
So, this is the type of automation that happens behind the scenes where a PM, for example clicks a button and then something happens and the system comes back with some sort of error. Now, do you know if our industry has implemented robotic process automation or RPA for sure to automate mundane project management or vendor management activities?
I’m not familiar with this expression, though I don’t think it’s very common in our industry. It’s more common in in manufacturing industries, if I’m not mistaken and so would you mind maybe explaining it a bit more, what kind of tasks do you have in mind that could be performed by this sort of process automation?
For example, you can teach it to check my email, check new orders from customers, if any email looks like this it’s an order from a customer, read it and log it into this ERP system that I have and from there I identify the language from the language. Obviously the ERP will give you a list of vendors and these are my top five vendors.
For example, sending email to these top five vendors for that specific language and by the time you come in the morning, most of the work is done for you. It’s as if the software replaced one of the project managers who would be doing all of this, clicking back and forth.
I haven’t implemented this, but I have a deep interest in RPA. Do you think that it has a place for automation in our industry?
Definitely has. Like any industry in these days in our industry is not secure from automation. In fact, it’s much needed because there’s still so much manual work going on. And yeah, but I think what you just mentioned, it definitely covers part of what we already do. It’s tier and even other companies do it successfully. And it can be extended even further.
So, one way to go about this is by building different integrations into different systems.
So one thing we do, for example, is we connect with our clients. Databases or ERP systems or e-commerce platforms or whatever they’re using to manage their content and when they require a certain type of translation, they, or even if they just like input a new product description, it would automatically be sent over API.
Integration to our system and our system would analyze it and depending on what the language combination is and what the type of content is, a different type of service would be ordered. In some cases you would go with a full on translation, editing and proofreading. In some cases you would go with something faster and cheaper because you don’t have such high requirements. And when we are done on our end with the process… So basically, when the translator is done with the project and the PM checks the file the project in Arc C Desk complete, it’s automatically sent back via API to the client’s platform formatted correctly and already available on their website or whatever they’re using. So this is it.
Like basic API integrations were not so basic. It gets kind of complex, when you go deeper into it. But we’re a lot of us in the industry and specifically as a tier we do a lot of this on a daily basis with a bunch of clients, and there’s more of this and the demand is definitely growing. So, this is something we’re seeing already, but but you were mentioning is something even more general.
It’s like, uh, including the you’re probably familiar with Zapier and similar tools that are very easy to use for day-to-day users. Something like that right? And integration with separate it was allowing you to set up a basic bots that’s going to do something depending on what you put the parameters sign, right?
So, I mentioned vendor management. Speaking of which, it is in an area where there’s a lot of repeat and mundane work which can and should be automated in my opinion. How do you see automation providing value in the supply chain process?
Well, one thing is it brings down the cost, so it style. We’re all about lean development. Don’t do costs that are not required so we can keep the prices competitive and we can keep growing in the marketand by keeping our costs down, we can keep the costs down for our clients. So, by automating a lot of our work we keep it more affordable to our clients and therefore everyone wins, right? And in the end if we don’t need 10 vendor managers but only need 2, to manage the same team of vendors and the same number of newly onboarded vendors.
It’s better for everyone and even our vendors are happier be given and their managers are happier because they can focus on better tasks they don’t have to do – click and repeat monkey jobs. They can do more interesting things and actually put their time into working with the people that they’re onboarding, right? So they develop relationships and all that.
So this I think the whole point of automation actually, not only in our industry, but in the in the global idea of automation as a whole was supposed to be not to take people jobs, but to make people jobs more interesting so we can work on the stuff where humans are actually needed and leave the stuff that’s not interesting getting to us to robots or computers or however you want to call them, and in vendor management, of course you can do a lot of things you can do testing partly automated. You can do data collection and sorting and inputting, partly automated and even all the way down to similarly what you do in digital marketing where you have fully automated campaigns and a lot of times users are not even aware of that.
You could even go as far and build relations with your vendors on that level it. I don’t prefer that, but technically it’s totally doable.
Absolutely. Now, Matija, we talked about the value that automation presents to the language company side of things. Now can you please talk about how automating things in vendor management and project management improve the experience for translators and interpreters you touched upon that briefly?, but can you elaborate a little bit?
One thing we doat Taia is that we keep everything in the same ecosystem. So, if you imagine, previously translators would need to either buy their own software or they would be provided with a copy of the software by the LSP that they’re working with and a lot of times vendors, translators, proofreaders and so on, are freelancers who work from home or from a coworking space or something like that, and for them, keeping up with the new tech and keeping their software updated and having ten different cat tools to use with ten different agencies that they work with, is probably a burden.
That’s one thing, and the other thing is, it’s also highly insecure, so when you as a client send your files via email to your local agency, and that agency sends those files via email to the local translator and the translator works in their pajamas on a 10-year-old laptop that they don’t regularly update and keep secure, probably in a public cafe or something like that, your files are not secure. Not anymore. Not at that state.
When we needed tires, we streamlined the whole process so everything the translators have to do is part of our platform. It’s inside the same ecosystem, so they don’t have to download the files. They don’t have to store them on their hard drives and they get to work in a cloud system in a cloud platform is always up to date and very fast and allows them to work much better than they would without decent cat tools or without any help from automation.
By our measurements they are usually capable of working up to three to four times faster than they would translate it manually. So, if you make a regular business document, which is something we do on a daily basis. And that’s not very creative.
Copy just the contractor, something that has to get done on a daily basis and has a lot of repetitive content we can with this level of automation with teams and all of that help them to translate much faster. So, for translators this is good because we pay them by the word. We don’t pay them by the hour and they can work much more efficiently this way in a system that’s always kept up to date for them and is always secure so they don’t have to burden themselves with security and all those issues.
This is one aspect, but there’s probably more.
Switching gears, a little material here. We’ve talked about the translation process and obviously what value automation can present on to both sides, the supply and the buyer side. Now automation and sales function has been a long running offering. I mean Salesforce and others have done a great job. How do you see sales automation creating value for translation companies?
That’s a very complex question, and one I’d like to say, see some answers to it as well. So, anyone out there listening to this, we would love some experienced people help us out with out marketing and our sales, because it’s not that we’re not doing well, but we can always do better and I love to learn new stuff, and here I think I’m at the end of my rope, when it comes to my skills for sales automation.
But let me walk you briefly thorugh what we do. We try to run marketing campaigns that are very targeted and we try to automate a lot of the lead generation process and all of that that’s going on and therefore deliver higher value leads directly to our sales team.
So, one thing we try to do is to keep our sales team occupied with warm leads and not get them to do cold calling so much and there that’s how they can convert better and work with people and invest their time into people who are actually interested in our product and are actually interested in changing their language, service provider or whatever that they’re using at this stage.
That said, it’s a very complex industry, it’s extremely fragmented. You’re going to find clients who have 5000 or more employees and don’t have a translation or localization system in place, even though they should.
But on the other hand, you might find very small startups who are very well educated when it comes to localization and have dedicated teams localization managers. They already have their software in place and for us it’s kind of one of the major issues at this point, and I’ll be honest about it, is finding the right type of client because we work with both very small companies and huge multinationals. We work with companies in very modern software as a service or tech industry or something like that.
But a lot of times we work with very traditional clients like manufacturing companies and so on. And it’s really like there’s no safe spot to say this is exactly our client, and that’s how we can automate the sales to get to them. This is something we’re quite honestly struggling with.
Let’s switch gears. Again, a little bit and talk about automating management processes and reporting analytics, if you will. This is an area I think that almost every industry is mature in terms of getting some degree of automation done. What are your thoughts?
First thing, you need to gather is your data. That’s that. That’s something we had to work around as well. So, even though we try to keep everything in the same platform, there’s still, you know, marketing runs in five different tools. Sales uses 4 different tools. Some other teams use other tools, so you need to usually what you would do is either find tools that can be connected well and go with something simple, like for example we use Google Data studio because it’s free. It’s easy to set up and it connects with a bunch of things. And it allows us to build interactive dashboards where we can find our all of our numbers on our KPIs. All important statistics in the blink of an eye for the enitre company and we’ve set this up in recent months and working very well for us.
But the next stage is definitely going to build a decent data warehouse to go with a proper PI tool that’s going to help you to analyze things much better. Because we’re in the era where you need to make decisions very, very carefully and selectively otherwise you can end up with wrong decisions and in this day and age with such a competitive and fast evolving market that can be very costly. So yeah, data driven. Make sure your data stored and prepared correctly and then go into analying it.
When it comes to reporting, I am a big fan of automated reporting. The one thing I hate most is when you’re working in a corporation and you spend half of your day preparing reports for your middle management senior management, that usually no one ever reads because they’re too busy making their own reports for their middle management, and so on and so forth up the chain.
So, this is something we definitely still can’t avoid at this stage, because we’re still a very small company, but I hope we can avoid it also in the future and keep of reporting and data gathering. Automate it as much as possible.
Besides productivity gains material, what else can automation offer to translation companies?
It’s not only about productivity and keeping costs and time required low. It’s also about increasing quality, right?
So, very old technologies back in 90s or even sooner, maybe already allowed for translation memories to be used and to help they keep things more consistent across the board.
So, if you have multiple translators working on similar projects, sharing a common translation memory can help them work more consistently and provide a more consistent quality translation. But in in recent years with the rise of AI and neural networks, what we see is also a huge potential in helping translators find better ways to express something or helping them keep the grammar correct and not only that, but also come up with creative new ways of saying things.
And by gathering all of this data with this new class of AI with transformer models, you can also generate a bunch of cool content. We’ve all seen last year what GPT-3 did a very impressive newcomer to the industry. But even from here on there’s already talk about way more sophisticated models that are going to help not only translate, but also generate content or recreate it, if you want re… What’s it called again? Transcreation, right?
This podcast is made possible with sponsorship from hybrid links, a human in the loop provider of translation and data collection services for healthcare, education, legal and government sectors. Visithybridlinks.com to learn more.
So that represents an opportunity for even new sub industries or subsectors to form or then the translation and localization industry talk to me about some of these. Where do you see automation playing a role? I mean, if we’re talking about neural nets or machine learning in general. What type of opportunities it presents for translation companies to reformulate what they’re doing and offer a business solution.
We’ve already seen this in the market in the last years. Some of the existing LSPs already started also working on data annotations services. So, when you have a bunch of data that someone needs to label and make sure that it can be used for training the new AI.
And since a lot of this data is usually informal texts and you have LSPs, on the other hand who already have an army of freelance series already working with them and they have all these processes in place, they are quite, for now, speed might be quite a good way to pivot into something like this, and we’ve seen cases of this in the last years.
But there’s, like you said yourself, with the growing industry and growing technology, there’s new niche products coming in that weren’t even available a few years ago and I’m really excited to see what else companies are going to come up with and what new business models can be evolved out of this.
How do you implement a vision that’s based on a combination of manpower and automation? Where does this thinking come from?
I think this there’s been a lot of talk and a lot of hatred going on against automation. In our industry specifically, but in other industries as well, and as I mentioned previously, I think the whole point of automation is to to make human lives more meaningful.
And we’ve been doing it since the Neolithic since we started settling down and becoming a civilized society. We’ve always found ways to make our work, our labor less intensive and made tools and made other things that help us and essentially I see AI’s and automation is just another tool that’s going to help us do things. It’s a very revolutionary tool, but so was the wheel, and so was the steam engine, and they made a lot of people to have to change their lives. But you can’t stop this. There’s no way you can you can stop this sort of technology from being developed.
If we’re not going to develop it, someonse else is going to. If people, who have moral standards or high moral standards are not the ones working on this, someone with lower standards is going to andthey’re going to rockhead work even worse, so we we need to embrace it.
We need to learn about it, not be afraid of it and find ways how to adapt our lives to work with it. There’s no other way arounf it, right?
And we’ve seen this going… This process has been faster and fast and afaster in the recent years and generations. But all in all, I think it’s we run on the brink of a very new era where a lot of our jobs are going to change dramatically and some for the better, some for the worse.
I guess some jobs that exist today are not even going to exist in 20 years. But it’s the same thing if you look 20 years back jobs from 20 years ago don’t exist anymore. No one fixes fax machines, right? They don’t exist, they’re out of date and we need to learn how to live with that, it’s just that thing.
So if there’s a piece of advice: Start learning, start researching. Don’t be afraid to dive deep into this technology, because it’s not as scary as it seems on the surface.
And for God’s sake, make your children go study computer science or something because we are in high demand of developers and engineers as a society in whole.
Absolutely, yeah. So just to add to your comment, I think as an industry we also have a responsibility to make people aware of what technology and automation is and that they shouldn’t be scared.
I mean some of our mega conferences, without naming any of them, keep talking about the low level implementation of, for example, different types of algorithms without telling people what business cases can be created from these new technologies and unfortunately everyone is talking about the how not not a lot of people are talking about the what like what can we sell, what these new possibilities available.
So, anyway that leads to my next question in order to implement automation, do you think it’s necessary to bring outside help in the form of consultants or other people who are familiar with the automation technologies?
Well, it really depends on the resources the company has, so a very large company is most likely going to look into acquiring a smaller competitor who has this kind of R&D development already established and has this kind of convergence kind of engineering already available. A much smaller company probably can’t even afford building their own technologies, so they’re going to have to outsource it to a consultant and outsource it to existing software in the market. And somewhere in the middle.
You can most likely go into developing partially your own custom solutions, but those can be extremely expensive. And yeah, going with the help of some advisors in our case, we obviously our goal is to develop technology that we can then use and also sell onwards.
So, we have our own development team. We work very closely on doing this ourselves. And keeping the knowledge in house, but when necessary, we’re not afraid of engaging with advisors who are much more experienced than we are, and might even bring us a lot more new knowledge to to to our games.
So yeah, there’s no best way to do it. You know, it really depends on what the issue is and what are your resources.
Let’s let’s take a look at our industry from the outside. Are there examples of organizations that you know of in our industry that have successfully adopted automation and if so, in your opinion, what type of gains are they reporting?
I think there are a lot of them. I mean there’s so much going on in our industry in recent years. If you just look at the number of mergers and acquisitions in the last year when it was a COVID year and I know a lot of things were stuck because of all the lock downs, but there were over, I think €8 billion of worth of transactions and about 39 different mergers and acquisitions going on. So this definitely shows that the market is very very vibrant when it comes to acquiring new technology our acquiring competitors.
So, I think what’s happening in a list in my opinion is the older players who have been on the market for a longer period of time and are very big and well spread. Don’t have such a good insight into what the technology is capable of or haven’t been up to date. Haven’t been evolving as fast as they should to keep up with the recent technological developments. So how they’re compensating for is that they’re acquiring younger companies who are more technically versatile and who knew how to handle all these new technologies out there.
So, it’s kind of an opportunity. On the one hand, big players can stay up to date by acquiring younger players who have this technology, and for younger players it’s a good way to consider as an exit strategy, I guess.
We talked about this in general terms earlier, but what about artificial intelligence and automation? I mean, you can automate things using rules and heuristics, but AI completely introduces a new paradigm. Here I’m guessing machine learning can introduce a different dimension and the automation concept altogether. How do we prepare or even pioneer such solutions?
In a large way, we already do it, right? So, the one thing that we’ve mentioned previously is the shift from statistical machine translation engines to neural machine translation engines, whereas like it only happened in the last five years, they didn’t even exist back before 2016. Neural machine translation wasn’t even a thing.
And now, I mean if you remeber, Google Translate was a joke to a lot of people and sometimes still is, but the quality of engines like that – this is just the most famous one, or infamous, however you want it – it has grown tremendously and I think what’s happening is that since the content is growing so fast, there’s loads and loads of new content being generated on a daily basis, and with the availability of localizing this content there will be the demand of localizing it, because it’s cheaper that any time in the past to translate it.
So, what’s going to happen is that even though automation is going to take over a lot of the localizatioon, it’s going to take over the parts that wouldn’t even get localized before. The parts that didn’t even exist, like new content that’s popping up over the internet on a daily basis.
And stuff that was traditionally translated by LSPs and other providers, is still going to get sold. Let’s say legal documents; you will probably not going to leave Google Translate to translate your entire mortgage contract with your bank for your house, right? You probably want to have that translated correct or whatever it is.
Whatever legal document you need, same goes with the user manuals, for example. If you are producing a forklift and your translations are done with Google Translate and that forklift is going to murder someone on some work place, then you will get into trouble, so that’s where you will invest into human in the loop sort of translation.
And this where we come in. So, we provide both. We allow our clients to select themselves and this is the part I love the most. Because in in the old days, the agency would be the one telling you what kind of quality you need. It’s the other way around. We let the clients decide what kind of quality they need. In what their budget allows for? And this is how automation then steps into all parts of the game.
You can go full on automate it, if that’s what you need, or you can go partly automated, but still ith a human in the loop process and go with that, and for me that’s the way it should be.
Let’s talk about some of the negative implications that automation presents or causes. We know that automation can reduce manual labor because obviously it’s taking away all that repetitive type of work. What else can it affect? Do you think that there are implications with regards to quality with regards to acceptance and any other areas?
Definitely. I think one of the main issues we’re going to get into, and we’re already seeing it is over reliance on automation. As automation gets good and better and better, we tend to rely more often on it, right?
I remember a very common case was with self driving cars when, uh, I think it was Google who had one of the first self driving cars and they gave it to a bunch of people and told them to use it. But they need to keep their hands on the wheel. At all times, similarly, what you have to do in your Tesla still right? And they put cameras in the car, as well. And in the first few hours, people were still holding hands on the steering wheel, because they were not sure if the car is going to do the right thing. But after a day or two driving around in a car that does everything perfectly and never kills you, you sit down and you relax and now you have all these videos on the Internet of people using Tesla’s ,who literally sleep in their driver seat and let the car do everything for them.
And that is a problem because technology doesn’t matter which level we’re talking about, either it’s self-driving cars that might get very very dangerous and kill people, or if it’s translation services. It can always be flawed, just like humans can always be flawed, so is technology. We should not over-rely on some things, especially in shuch early stages of this technology and we see similar things with translations.
When we measure the distance our translators do, the better the machine translation output is – the lower the distance that they do when they translate something, so they they get presented with a pre-translated piece of content,which they compare with the original segment and if it’s more or less correct, they’re less inclined to fix it because it’s more or less correct.
And we can see the level of engagement on the human side then strats dropping. It’s very easy to let a mistake slip this way, because you don’t see it, because you rely so much on the MT, you just go OK, this is fine, move on to the next segment. This looks fine, move on to the next segment and then you just skip it. You just don’t see.
But there’s an error somewhere in the middle, right? So overreliance is going to be a problem that we’re going to have to work around still.
Well, I agree with you that we are quite some distance away from level 5 type of automation. We are still dealing with augmentation with humans. Now about implementing automation at organizations. Do they affect us on, the client side, of course. How do we respond to the need for bbeing faster and more efficient as expected by algorithms implemented on our client side?
That’s a very complex question. Uh, yeah there’s I think there’s only one way of going about it, so the time is here to automate or to be left behind. I think that the players in the industry who will adopt new technologies and will follow the trends and will learn and and develop will be the ones surfing the new wave of automation and the ones who won’t will be left behind and the tsunami that’s coming is going to sweep them out.
So, there’s no way around it.
Just to reformulate that question. Let’s say if your client that has been sending you work for manual translation all of a sudden implement their own algorithms where as a supplier of translation services for example, or even in some cases, interpreting where would you add value now.
Because the way I see it, for example, if your client has some sort of a streaming service, and before that you had to translate the subtitles and now they’ve automated all of. That, uh, I see an opportunity where you can actually do a handoff of automated transfer.
To your company when the content is dealing with something very specialized and you can just add value by correcting it, and by augmenting what their algorithm has done. How do we go about doing that?
Where does an LSP pivot to or how do they pivot in order to make sure that they deliver the right type of value?
So, an LSP is quite definitely going to keep being the human in the loop part. So, if your client is let’s, like you said your example, you have a video content producer or a company that deals with that they need a lot of localization. Previously, everything was done manually. All the subtitling. Now the transcriptions and a lot of the subtitling is already fully automated. But it’s not perfect. There’s still a need for the human group, at least for the next five years we see it will still be there.
And depending on the type of content and the type of users who will engage with this content, probably even much further into the future. So, if it’s a YouTube video, most people are perfectly fine with the with the automated CC right, but there’s still some content where humans will always be present in the loop, so that’s where analysis we can place themselves.
The problem which I see here is, the one that you mentioned, if a client were to do this on their end – how do you then integrate with their system to make sure that you’re compliant on your system?
I look at it from the technical perspective because our, at Taia, at our company, our paradigm is that we are the ones who provide the technology, we’re the ones who provide the platform for our users. We are capable of integrating with a bunch of different systems, but we rather keep everything on our end and get the clients to use our platform.
So yeah, if this were to happen to us, for example, if one of our clients and we do subtitling, ofcourse as well, we were to decide to automate it on their end entirely and we were just a part of their loop, we could engage with them via an API and keep providing our human in the loop services. But, we rather see they use our platform, so we can keep everything on our end. It’s easier for everyone.
Yeah, that that would be an ideal scenario, but let’s talk about, uh, where do we go from here? How does the next five year look like in embracing automation within localization industry?
Cool, that’s $1,000,000 question. Or even more. I’d love to give you an answer to that, but unfortunately I don’t think I’m well experienced enough for something like that.
So definitely what’s going on is the automation is going to take off a lot of work. Users or companies who didn’t use an MT aredefnitely going to satr using it. New jobs are going to pop up, as we already mentioned.
And I think the whole industry is about to get shook over and and swept over by this tsunami of automation.
What the end game will be? What will be the result of this?
I think a lot of the players are going to go extinct, there will be even more mergers and acquisitions coming in. A lot of the market will probably consolidate because it’s a very mature market, but it’s extremely fragmented, so it will probably start consolidating even more.
And we’re already seeing this trend in last years and whoever is not going to automate is just going to be left behind. That’s basically the point of this. But we’re playing it.
So, so that being said, do you think our industry is prepared to adopt automation at the scale that we are about to see?
I don’t think any industry is prepared to see what what’s coming in the next five years, when it comes to automation.
As I mentioned previously, just remember when GPT came out, even GPT 2, can you remember the type of content it was able to generate and then number three was even more and you know that we can’t even predict what would what AI will be able to do for us in five years.
It’s too far ahead for people who, like myself, who are not very, very specialized in this research. We’re merely developers and consumers. We’re not researchers. And there’s quite a bit of a gap there.
I don’t think any industry is prepared for what’s going to come up. It’s we’re all going to have to embrace it or fail, and I think some players are much more well versed in this technology and will be able to support their clients needs and support their their growth by using this technology. But not everyone is going to be capable of doing this, obviously.
So, yeah, indsutry as a whole, definitely not some players, sure. Soure, yeah, definitely, yes.
Matija, what’s your message for our industry and your peers? What would you like to tell them about automation?
Well, I think we’ve talked a lot about automation. We talked about how important it is to embrace the technology, how try not to be afraid of it outright? Let’s try to work with it, rather than against it.
But I think in the end my message should be that it’s not all about automation, it doesn’t matter what kind of technology your company has, if you don’t know how to work with your people. If you don’t know how to work with your clients and present a decent human connection, there’s no way you can keep your company successful or make it successful.
It’s, I think with all the automation coming in, it’s becoming even more important to focus on the human aspect of whatever business that we’re working on, and I honestly hope, I sincerely hope, that our generation will be able to, on the one hand, embrace new technologies, like AI, to help us work faster and more efficiently, but on the other hand, to also live more meaningful lives that will allow us to go in for deeper human connection and work ona more, we call it culture in our company, what we’re developing, but I think it’s more of a mentality, even if it’s more of a mindset.
So, you can find time to work with people and develop this deep relationship with them. That’s my idea here. So, hopefully we’re going into this direction.
The other the other idea is just we’re going into a dystopia, but that would be more…
I hope we’re going to get there. So, Matija, I agree with you, automation is the future in so many ways, and, I think our industry is also jumping on this bandwagon and we must stay on top of the latest trends in this area so hopefully we can revisit this conversation again.
It was a pleasure speaking with you today and as I said, I’m hoping we can do this again in the future and have you in another episode where you can talk about automation and what’s happening with that.
I want to thank you for your time.
Thank you, Sultan, forhaving me. This was a very engaging conversation and I hope we get to talk again soon.
That was a very interesting conversation with Matija. I think he represents what our industry is undergoing at an early stage at the moment.
As you heard, automation is perfect for offloading repetitive and mundane tasks from our human workforce to computer processors while freeing up our people to do more exciting things.
You should ask yourself why should your project manager spend his or her entire day copy pasting when you could do that with an automation solution.
Your project manager instead could be spending time talking to your client providing the human touch and making them feel good about doing business with your organization.
Automation does not necessarily mean loss of jobs. It simply mean the balancing of work. I think the reason why computers are good with automation is because they lack creativity, which is something neither for more exciting and important things. And you people will enjoy working with people performing more challenging and unique tasks.
This brings us to the end of this episode of the podcast. I had a lot of fun talking to Matija and learning from his experiences.
I hope you also had a few action items that you could take and apply to your processes in your organization. If you found this conversation helpful, then you don’t want to miss our future episodes.
Don’t forget to subscribe to the translation Company Talk podcast on Apple Podcasts, iTunes, Google Podcasts, Spotify or your platform of choice.
Thank you for listening. Make sure to subscribe and stay tuned for our next episode.