Janus AI: a Game-Changing Plugin for Client Projects

In early 2024, the Janus Worldwide team launched a large-scale initiative to implement its proprietary AI plugin in real-world production scenarios. The approach was practical and business-oriented: how to reduce translation costs without sacrificing quality, especially in the face of limited budgets and strict client deadlines.

The project started with pilot launches and internal testing. Since July 2024, the plugin has been actively used in production tasks. At first, it was without glossaries, and later with full terminological support. This phased launch allowed us to flexibly adapt the system to clients’ requirements and build a stable workflow.

From TEP to MTPE: a new production model

A key decision was the exclusion of the traditional translation stage (T) from the TEP chain. Instead, a process was introduced in the Janus AI plugin, which generates a primary translation of sufficient quality for subsequent editing (MTPE) and final proofreading.

The results were not long in coming:

• RU–EN—minimal revisions, high quality, and fast delivery.
• EN–RU—reliable quality and low editing volume.

Financial efficiency and scale

The economic effect was impressive:

• Translation costs were reduced by 28% without compromising quality.
• Over the course of a year, the plugin was used to process more than 1,300 projects.
• The total volume of translated content was over 4 million words.

And, most importantly, the plugin itself has evolved significantly over this time. For example, if at the start the translation of 10,000 words took several hours, today this job is achieved many times faster.

How is quality ensured?

The transition to automation does not eliminate high responsibility for the results. At Janus, the Quality Assurance process is built on a clear methodology:

1. Analysis of client requirements: guidelines, stylistic instructions and evaluation criteria are studied.
2. Creating design specifications: instructions for all participants in the process—editor, translator, proofreader.
3. Setting up automatic checks: connecting QA profiles in CAT tools.
4. Terminology management: working with client glossaries or creating new ones from scratch.

LQA as digital quality control

The final stage is quality control. This is where it comes into play LQA (Linguistic Quality Assessment), an expert review during which the following are assessed:

• the degree to which the translation meets the client’s expectations;
• number and type of errors.

Assessment is carried out by internal and external specialists, ensuring an objective metric. This allows us to not only monitor quality on a regular basis, but also to make management decisions based on the figures.

What does this give to the client?

1. Accelerated processes without compromising quality.
2. Controlled quality confirmed by an audit and figures.
3. Scalability: the plugin works successfully in projects of varying size and complexity.

Sometimes the boldest ideas come not in the “laboratory”, but from everyday practice. Our hero, Artyom Burov, a key account manager in the Life Sciences department, proposed a concept that revolutionized traditional processes and opened up new work scenarios. In this interview, we explore how the idea was born at the intersection of real-world client challenges and personal experience, and what steps were needed to transform an internal initiative into a product with potential.

• Artyom, please tell us how the development of the Janus AI plugin began. What was the starting point?

The story began with the appearance of a new client with very low rates. It was obvious that it was impossible to work in standard mode under such conditions. With the support of CEO Konstantin Josseliani, Vice President Igor Mogilevsky, and Department Head Marina Shirshova, I began looking for ways to not only maintain but also increase the projects’ profitability.

At first, I tried using machine translation: the R&D department developed a custom engine for the client based on Modern MT, but the quality of the results was unsatisfactory.

At that point, we met Valeriy Bolshakov, who suggested we try creating an AI-powered plugin. The idea immediately seemed promising, especially since the use of AI was already being actively discussed in the industry. Then we realized that it was time to launch our own initiative.

• Which improvements had the biggest impact on performance between July 2024 and fall 2025?

Initially, the plugin only worked with memoQ, and the translation process was quite complex: it was necessary to create an offline version of the project, export the translations to RTF, and upload them to the server to connect translators and proofreaders. Now, a fully-fledged infrastructure has emerged that allows you to work with the plugin directly in the working version of the project, significantly speeding up and simplifying the process. In addition, colleagues developed a version of the plugin for Trados, which was an important step in its development and expansion of capabilities.

• Can you remember the moment when the team first saw real business results from the plugin?

Yes, the moment when the team saw real results was clear. Linguists began to notice an improvement in the quality of translations, and for us, the effect was immediately obvious in the numbers. Project margins increased by 28%, and it was hard to miss.

It then became crystal clear to us that the translation stage could be accomplished using a plugin. Since then, we have been working according to the following scheme: translation using the Janus AI Plugin, then MTPE and proofreading.

Special thanks to our colleagues in the QA department, as they help generate reports on all projects quickly, allowing us to promptly identify and correct any potential issues.

• How do you assess the long-term role of the Janus AI plugin in transforming the TEP model at the company level?

I believe the plugin can truly transform processes within a company. In the long term, it could completely replace the translation stage. In our department, we are already noticing that, given the reduction in rates, more and more projects are being processed using the plugin. In my opinion, the future belongs to AI-powered translations, and this trend is confirmed by leading market experts.

We clearly see how the plugin helps increase efficiency and reduce labor costs. However, human expertise remains necessary to ensure high quality translation.

• If you were to start the project over again, what would you do differently at the launch?

To be honest, I wouldn’t change anything from the production side. We developed the plugin gradually, step by step, carefully refining the details. This approach allowed us to achieve a sustainable and high-quality result without any sudden movements.

• What was the most unexpected thing you experienced while working on the plugin?

I was surprised by how quickly artificial intelligence began to change our industry. We all understood that it was a matter of time, but frankly, I didn’t expect the changes to come so quickly. Work on the plugin clearly demonstrated that technology is already capable of transforming familiar processes.

Janus AI is not just a technology solution, but a strategic tool for teams working with limited budgets but high demands. Thanks to a thoughtful approach to quality and full integration into workflows, the plugin becomes a real assistant in achieving the business goals of Janus Worldwide clients.