Argos MosAIQ Introduction Video with Erik Vogt, Innovation and Solutions Director

Argos MosAIQ is a translation platform that combines AI technology with human expertise. It speeds up translation work, lowers costs, and maintains high quality standards. The platform uses client terminology, style guides, and industry knowledge in real-time, along with detailed analytics to deliver customized translations.

The platform’s main component is the Mosaiq Editor, which manages AI tools throughout the translation process. These tools check source content, improve translation memory matches, enhance machine translations, and perform quality checks using MQM standards. This reduces manual work while ensuring consistent, high-quality results.

Mosaiq helps translators work more effectively. Rather than replacing human translators, it handles routine tasks so they can focus on creative and strategic work. This combination of AI and human skills increases productivity by two to three times compared to standard methods.

The platform adapts to different client requirements with features like content optimization, smart MT engine selection, and translation memory management. Companies using Mosaiq improve their translation processes, save time and money, while maintaining excellent quality.

Learn more about how Argos MosAIQ is transforming localization workflows in this introduction by Erik Vogt, our Innovation and Solutions Director.

Stephanie Harris-Yee: My name is Stephanie and I’m here with our Solutions and Innovations Director, Erik Vogt, to talk a little bit about Argos’s new AI translation solution, Mosaiq. So, Erik, can you give us a brief overview of what the product is and what it aims to accomplish?

Erik Vogt: Sure. Thanks, Steph. Yeah. Mosaiq is an application of a lot of the, best ideas in the AI space. I think we’re focusing on the human-in-the-loop element. So it’s largely based on helping the translator have the best possible support. that often means doing a variety of enhancements. Now, one of the things that I think the industry generally gravitates around is that AI is very good, but it’s also even better if it’s focused in specific ways. So we have a multi-agent model where each agent is responsible for different specific things. And we aggregate those to build a total project Mosaiq, which includes a whole bunch of different capabilities. So some of those capabilities we think of is dynamic prompting. It’s our core entity that takes client assets, which are glossary, style guides, source materials, TMs, and provides those as resources and then creates prompts that are specific to a targeted purpose. So for example, improving terminology or improving style, and then the AI makes recommendations based on those two, whatever it’s starting with. So whether it’s translating from scratch, which is one of the options, or taking either a TM output or an MT output, and then getting it the rest of the way there. So some of the key features include fuzzy match enhancement. That’s a key entity to really make the sort of what it’s starting with, get it closer to final for the translator. When it’s making QA statements, and quality assurance is a huge part of our mindset on this, it can do things like make recommendations based on the style guides or glossaries. And one of the nice things about what AI can do for us is not only make the recommendation, but explain why, and it’ll give some insights into why this is a better solution for us. It’s also highly customizable. So because we’re working on an agent-based model, we can apply these agents in very specific ways that are appropriate for each particular use case.

Stephanie Harris-Yee: Can you share any case studies or examples of how it’s already been used successfully?

Erik Vogt: Yeah, sure. One of our use cases that is particularly exciting was an e-commerce platform. They were struggling with returns due to unclear instructions. And so when you’re dealing with the international market, returns are very expensive, especially expensive. So, we sought to improve the instructions using Mosaiq. And we were actually able to deliver up to 50 percent savings on the number of returns that are required just due to the quality in the product descriptions alone. There’s another really great use case for AI which essentially has a creative element to it. And this is where we have AI help out with blogs. Blogs are marketing content that really benefits from a local spin. And that’s something that AI is really able to do. In this case, we used one of the features of Mosaiq, which is being able to measure productivity on a per segment basis. So we can really look at how the translator is working with the content. And  measure how efficiently they’re doing it.

And in this case, in some cases we actually have productivity over 300 percent faster than the, process that we started with. Basically delivering over a thousand words an hour. So this is capability that leads Argos to, to deliver a much higher throughputs and there’s quality savings there.

And again, it’s not always that fast, but this is something that’s available. But it also provides the quality support, as I mentioned earlier, which is one of the things we don’t want to do is rush translators into getting things done faster because there’s a huge risk of of creating negative incentives there. But the quality support tools that are there give us an opportunity to add fluency and quality. On top of an already high-quality bar. So I think this is really, amazing stuff. AI also has a great use case in translating user-generated content. It’s excellent at interpreting a variety of inputs and being able to normalize those according to standards and then translate them.

Stephanie Harris-Yee: Well, it sounds like a very, very cool tool. Are there future updates or features in the pipeline for Mosaiq?

Erik Vogt: Yeah. I think on top of the enhancements that we benefit from just using AI in general, and I think we’re not limiting ourselves to one AI tool out there. We can switch and enhance them, our process built on top of the capabilities that that the large, large AI foundation models deliver. So as they get better. We get better. In addition, we also plan enhanced LQA methodologies. I think that’s a critical one. We’re adding additional languages. As we all know, AI works much better with some languages than others. So some language pairs we’ll be adding as they reach the target quality levels that we’re expecting. Integration is always a big one. I don’t think there’s any end to potential integration capabilities. I think we all know how challenging that could be, but we will be enhancing the capabilities of Mosaiq through additional integrations, efficiencies, and automations. And lastly, it’s all about the data.

So as I often say, whoever has the best data wins. I mean, understanding what you’re dealing with, both in terms of throughput, yeah. in terms of quality, in terms of efficiency, in terms of any number of metrics that you value in terms of your overall engagement with your audiences or with your content or compliance standards, all these things will be delivered through an increasingly granular and efficient analytics dashboard. So as we continue to develop the product, increased LQA capabilities, better language support or broader language support, but more integrations and the analytics dashboard are key features.

Stephanie Harris-Yee: Wow. All right. Well, now here’s a question that always comes up when AI is concerned, but what steps have been taken to ensure that Mosaiq is secure?

Erik Vogt: Yeah, that’s a good one. You know, I think data encryption is a big part of it. So our process is encrypted end to end from the… every authenticated user and we can control the access roles and how much they have. So our client’s content is never out of control. It’s encrypted both in route as well as at rest. So the access controls the data encryption… Argos is an ISO-compliant, EU-centric, language provider. So naturally, GDPR is not a concern. We absolutely comply with all GDPR standards and there’s regular audits that go along with that by independent entities that ensure that all this is being done according to our stated policy. There’s something even more thought-provoking here with regards to the AI. And that’s the importance of the retention. How much are you letting the AI remember? We are strict in the sense that we do not, give AI any data. The AI processes the data and then forgets it. So it is never stored in any nebulous cloud anywhere that the data is controlled through the assets that we

share during the translation process, and once it’s done, there’s no storage by the AI after that point. So this gives us end-to-end control and again synchronizes with our very targeted strategy of adding value through these individual agents in a controlled process. That is 100 percent encrypted, GDPR compliant, and audited.

Stephanie Harris-Yee: So if I’m a client and I want to test this tool, see what’s going on, get started today, what should I do?

Erik Vogt: Well, the clients that we’ve talked to so far have really appreciated the fact that we’re not just talking about this. We can actually show it. So the best possible thing to do is to set up a demo. Providing some sample content and some assets, such as your TM, glossary, and style guide will help us for,

For give you a demo that is, as targeted to your particular needs as we possibly can. Pilots are a great way of showcasing the full capabilities of it.

But even a light demo, if you’re not quite ready to do that can really give us an opportunity to show you in real time, just how exciting this capability is.

Stephanie Harris-Yee: Wonderful. Okay. Well, thanks so much, Erik.

Erik Vogt: You’re very welcome Steph. Thank you.