Translating Podcasts and Videos in Your Own Voice Now Possible Thanks to Translated
Translated’s new voice translation service seamlessly blends transcription, machine translation, expert linguists, and voice cloning technologies.
→ Continue Readingprofile
Mapping Milestones in the AI Revolution
Three CEOs weigh in on the past, present, and future of language technology
By MultiLingual Staff
I
t didn’t take long for the global artificial intelligence (AI) arms race to heat up in 2025. In January, Chinese lab DeepSeek shook up the industry with the release of the R1 large language model (LLM), which reshaped expectations for what the technology could achieve at low costs and computing power. The year only heated up from there as established AI giants released their own new and improved models.
For the language industry, the business landscape was no less dramatic. Companies like Unbabel, Translated, and DeepL worked steadily over the past 12 months to deliver more sophisticated language technologies capable of increasingly accurate translations. And with the industry seemingly evolving from week to week, who knows what the future might hold?
The quick answer to that question: no one. But some have a better idea than others. With that in mind, we asked Unbabel chief executive officer (CEO) Vasco Pedro, Translated CEO Marco Trombetti, and DeepL CEO Jarek Kutylowski to map out the latest in the AI revolution. Fortunately, they found time in their busy schedules to share their insights.
Advertisement
UNBABEL
Vasco Pedro
What, in your opinion, are some of the most exciting or impactful dynamics happening with AI vis-à-vis language technology and agents?
What we’re seeing right now is a major evolution from AI as a language tool to AI as a language partner. The real shift is not just comprehension but autonomy. AI systems are starting to reason, initiate, and act independently. That’s where things get really interesting.
A few key dynamics include:
- AI agents: We’re seeing the emergence of agents that initiate, reason, and execute. These systems are capable of navigating ambiguity and orchestrating complex workflows across multiple platforms. This unlocks a new class of applications: autonomous customer support, multilingual market research, proactive content generation, even negotiation. It won’t be long until we get to a place where agents develop the generalized ability to act across domains.
- Natural language understanding (NLU): With models like GPT-4, BERT, and other advanced transformers, we’re seeing AI achieve human-like understanding of context, sentiment, and nuances in language. This enables more sophisticated conversational agents that can adapt to a wider range of user inputs with higher accuracy. In high-stakes fields like healthcare, law, and finance, performance alone isn’t enough. What matters is why the AI made a decision. Explainability and traceability are becoming just as critical as accuracy.
- Multilingual capabilities: AI systems are increasingly able to work with multiple languages at once, enabling real-time translation and communication across linguistic barriers. This has vast potential for global business operations, accessibility, and education.
- Personalization: Language models are being fine-tuned to create more personalized experiences, from tailored chatbots to dynamic content generation. AI can now analyze and adapt to individual user preferences, providing highly relevant and effective interactions.
- Integration with other technologies: AI is being integrated with tools like voice recognition, vision systems (useful for applications like generating captions or interpreting sign language), and even augmented reality. The synergy between these technologies can create powerful multimodal language agents that can serve a wider variety of use cases, from virtual assistants to immersive learning experiences.
What is your vision for how these technologies will transform the language business and its day-to-day operations?
AI is fundamentally reshaping the language business — from workflows to workforce. In many respects, it’s redefining how value is created, and by whom. Some of the key developments we’re seeing include:
- Automation of repetitive tasks: For translation, transcription, and content creation, AI can handle routine tasks, freeing up human professionals to focus on more complex or creative aspects of the work. This could increase productivity across industries like legal, medical, and business communications.
- Enhanced collaboration: AI-powered assistants could facilitate real-time collaboration between individuals from different linguistic backgrounds, reducing the friction of language barriers in meetings, writing, and collaboration tools.
- Improved accessibility: In fields like education and customer service, AI could make language tools more accessible to nonexperts. Real-time translation, voice-to-text systems, and AI-powered learning tools could support more inclusive learning environments and global collaboration.
- Business models and cost efficiency: Language services may evolve to leverage AI for initial drafts, translations, and content generation. Human professionals would then focus on editing, refining, and adding cultural or contextual nuances that AI might miss. This would likely reduce costs and speed up workflows, as well as create a need for professionals who can effectively manage and oversee AI-driven processes.
What should language professionals expect, for instance, in terms of division of labor?
The key takeaway should be that AI is not here to replace language professionals but to augment and empower their work. AI language technologies are tools that, when integrated thoughtfully, can streamline operations, improve accessibility, and create new opportunities for innovation and collaboration.
So, this isn’t a case of man or machine but rather man with machine. Ultimately, language professionals who embrace AI as a co-pilot will find themselves doing higher-impact work, at greater scale, with more strategic value.
The future belongs to those who can steer the machine.
How do you expect the conversation around AI to evolve as the technology progresses?
Things to keep in mind as this conversation evolves:
- Bias and trust are fundamentals: As AI becomes more integrated into language services, issues related to bias in language models, data privacy, and fairness will become more important. Ensuring that AI is used responsibly and inclusively will be crucial for the future of the industry.
- Human-AI collaboration: It’s important to highlight how the best outcomes will likely emerge from collaboration between AI and human professionals. For example, AI can handle language processing at scale, while human insight can refine meaning and ensure cultural sensitivity and nuance.
- Collaboration is key: AI models are not static. As they continue to learn and evolve, the expectations for language technology will change rapidly. Businesses and professionals in the language industry will need to keep up with these shifts to stay competitive.
- Generic AI won’t cut it: AI tools for language processing will need to be increasingly tailored to specific industries, whether it’s healthcare, law, or entertainment. The more personalized these tools become, the more value they’ll deliver in helping businesses operate more efficiently.
TRANSLATED
Marco Trombetti
Where do you think we are now in terms of AI technology development? We’ve seen its capabilities become increasingly sophisticated, so where does that put human workers, particularly language professionals?
We’re entering the “autonomy era.” Before 2007, AI in the language industry only slowed translators down. Then it began boosting productivity and improving translation quality. Now, for a limited number of languages with enough high-quality data to enable reasoning and support agents, we’re entering a phase where our role as humans shifts to guiding and supervising the work of machines.
As technology evolves, so does human potential. And as human potential grows, so does technology. At the heart of this dynamic cycle lies our belief in humans, the true drivers of innovation. By embracing a new symbiosis between human creativity and AI, we amplify our collective ability to shape an inspiring, limitless future.
Today, we’re witnessing a huge, latent demand for translation. Soon, we’ll be able to translate orders of magnitude more content and finally allow everyone to understand and be understood in their own language. We need to unlock this value as soon as possible. The world needs more understanding: in business, in politics, and among people.
You mention the latent demand for translation. What does that mean for companies seeking language services and for the language industry itself?
Companies will grow internationally faster, with the capacity to deliver content that precisely aligns with user needs. It’s a new and different domain of translation from what we know. It’s no longer just translation between languages; it’s content adaptation to match users’ level of education, available time, and interest in the subject. People will soon experience the entire web in their language and their preferred form, connecting with everyone else on the planet.
On a day-to-day basis, we will have to work hard to open up this opportunity and meet these expectations. This is a very demanding task, and many will struggle because they see it as too complex. Some will give up under the pressure of too much work and uncertainty. But others will never give up.
As the language industry continues to adapt to often-dramatic changes, what is the message that professionals should take away?
This is not the time to give up. This is the time to double down.
How do you view the nature of this change? Are we looking at procedural adjustments or a fundamental reordering of the industry?
This is not a moment of transition; it’s a shift. A cultural, operational, and emotional shift. We are moving from a world where AI assists humans to one where humans guide autonomous systems. Translation is unlocking the next level of communication, not only among humans but also between humans and machines.
How has Translated managed its approach to these rapid developments? What are some of the latest innovations and exciting developments from the past year?
Translated has been dedicated to providing translation AI and professional translation services to enterprises. By delivering symbiotic translations, powered by humans and machines working together, we’ve collected the data needed to train an AI that outperforms what’s commercially available today. Recently, we made this technology accessible to everyone on the web.
Lara is the result of 25 years of hard work. It can translate text and documents, serve as a consecutive interpreter, and now even function as an independent AI agent. It’s compatible with Modern Context Protocol (MCP) and can make any LLM multilingual, integrating translation into any AI-driven process.
Did your team secure any new funding in 2024? What has historically driven investor interest in Translated? What sets Translated apart from other companies in the market?
Translated has always been a highly profitable and efficient company. This solidity has allowed us to cover the significant investments required to deliver the widest-reaching language support and the most reliable translation AI without the need to look for external funding.
Can you share more about your growing customer base? What types of customers do you work with, and can you provide any specific use cases?
Translated had good growth of 22% this year. We love to partner with visionary and forward-thinking brands, and we recently signed a large deal with a brand we consider one of the coolest in the world.
Advertisement
DEEPL
Jarek Kutylowski
With so much activity in the AI space over the past year, what are some of the major milestones for DeepL?
Last year was a big one for DeepL. In November, we launched our first product that translates spoken language: DeepL Voice.
This was a major milestone for us — live speech translation is really the next frontier for DeepL as a business. We’ve concentrated on the written word for so many years, and the quality has exploded since 2017. Now we’re going through the same process for spoken language, making it possible for companies to translate virtual meetings in real time. DeepL Voice also powers spoken conversations outside of meetings and can be used for one-on-one conversations just using your phone, helping our customers with so many new use cases.
Another thing that I’m excited about is a feature that we just recently launched this year called Clarify. This is a new addition to our translation solution, and it allows you to more deeply interact with the AI when you’re going through the translation process.
Think about all of those cases in which language is ambiguous — such as words with multiple meanings, specific cultural references, and specialized terms. All of these ambiguities can impact the accuracy of translations, and when it comes to business communication, finding the perfect word or phrase is really important. With Clarify, our AI proactively participates in a dialogue with users, prompting them with questions to clarify context and address any potential ambiguities, making the translation outputs even better.
We think of Clarify as an interactive AI-powered language expert that is there to help you think through exactly how your words come across in different languages. This is a major milestone in the way users work with translation nowadays.
DeepL announced a new $300M investment in May 2024, led by Index Ventures, which brought DeepL’s valuation to $2 billion. What do you think drives this interest in DeepL? What sets DeepL apart from other billion-dollar companies in terms of attracting investments?
We’re pretty unique as a company in that we’ve merged the deep tech aspect — conducting cutting-edge AI research, working with the models, designing models, training world-class models, and so on — with an understanding on how to bring the technology to our customers — defining the applied aspects and creating the products and applications that are so necessary to leverage the power of AI tech.
There’s been a big shift in the overall AI industry from research-only prototypes to truly impactful products. And as a company, we can demonstrate our impact clearly, which our customers are happy about. We can also clearly calculate the return on investment with our solutions, something that, obviously, investors love.
Can you tell us more about your customer base? What type of customers do you work with? Are there any specific use cases you can provide?
DeepL really shines in high-value, business-critical use cases and is horizontal in terms of its application. Most industries are global nowadays. Today we work with over 200,000 businesses worldwide across multiple industries, from manufacturing and legal to retail, healthcare, and life sciences, including with leaders like SoftBank, Mazda, Harvard Business Publishing, the Ifo Institute, Panasonic Connect, and more. All of these businesses rely on DeepL to help them scale and collaborate more effectively across languages and markets.
Our platform is trusted both for its unprecedented quality and accuracy as well as from a data security perspective. We’re seeing customers who, for example, translate quarterly earning reports before those can be published. And this is obviously highly sensitive data, but it’s also data that is incredibly important when it comes to accuracy.
We’re also seeing companies from Asia, like Japanese car manufacturers, with research and development (R&D) locations in Asia and large cohorts of their customers outside of Asia. The direct link between the customer-facing organization and the R&D organization makes for better products. A lot depends on the quality of communication.
Where are the most exciting or impactful trends in AI, particularly regarding language technology and agents?
When it comes to what’s next and what will be most impactful when it comes to language AI, two things stand out to me. I’ve just spoken about live voice translation: This is a game changer for many people and for many companies.
The second is the evolution of language AI tools from a static, one-way experience to being much more personalized and collaborative — having the AI understand you as a person and you as a company. There’s enough generalized AI on the market, which gives you an average of the internet translation, basically. But what we’re offering is much more specialized and accurate, and when it comes to the future of our technology, we’re working very hard with our customers to get much more customization and context into the mix. DeepL was always strong in incorporating context into translation — this is part of where our high accuracy comes from. But we’re looking further into how we harness the data that the company has, for a more personalized, intuitive experience. How can we integrate this data into our AI models so that they can translate even better?
What do you think the future for DeepL looks like? One year from now, five years from now, what do you envision?
Five years is like 100 years in AI! But if I think about DeepL in one year, then it’s largely about this personalization aspect.
I think, in a year, DeepL will be able to understand each and every one of our customers much better, and in five years, I expect our AI solutions to be much more deeply embedded into user workflows and even more personalized and context aware. DeepL will understand you as a person, and your business, much better and will be able to deliver translations so accurate and natural that you basically won’t have to make any edits anymore. We’re excited for the future!
Advertisement
Related Articles
Unbabel Report Outlines “Native Language Effect” among Consumers
Unbabel, a San Francisco-based language operations platform, has released the results of its 2021 Global Multilingual Customer Experience Report, which breaks down the results of…
→ Continue ReadingAlibaba AI Real-Time Translation Expands to 119 Languages
Alibaba AI real-time translation now supports 119 languages, offering real-time interpretation, text, image, and document translation in one platform.
→ Continue Reading