Episode Transcript
[00:00:00] The Wellness Intelligent Automation and sustainable localization at WellHub by Josevi Abad Making every company a Wellness company that's the mission at WellHub, a corporate well being platform connecting employees to an extensive network of gyms, studios, personal trainers and wellness apps for fitness, mindfulness, online therapy, nutrition and sleep. It's not just empty words. It is an objective we really believe in and a principle that guides every decision we make. This declaration of principles served as a strong starting point when in the summer of 2025, our company handed down a new strategic approach.
[00:00:46] When a variety of organizations are exploring the benefits of artificial intelligence AI, we investigated how we too could implement the technology to do more with less. Our MKTL10N team is in charge of all localization operations for our marketing division, and this new approach did not come to us as a surprise.
[00:01:07] As localization professionals, we are well acquainted with technology developments.
[00:01:13] That's why on a daily basis, we use a wide range of productivity tools to manage our workflows.
[00:01:20] We likewise know that generative AI, gen AI and large language models are touted as the silver bullet for democratizing communication and slashing costs to near zero. But promises of instant free translation often conceal significant legal, environmental and quality compromises.
[00:01:42] Therefore, we face the dilemma. We could plug into an open gen AI model, essentially outsourcing our thinking to a black box.
[00:01:51] Or we could find a way to innovate in a way that would keep well being at the very core of everything we do. This is how we built a healthier operational ecosystem, internalizing our operations to achieve automation in an ethical way that prioritizes data sovereignty and chooses quality and sustainability over hype.
[00:02:11] The Ethical Supply Chain Problem the conscious choice of our localization team to deviate from a standard gen AI approach stemmed from the fact that the current gen AI generation comes with hidden costs, often glossed over in boardroom presentations.
[00:02:28] While a chatbot interface is clean and user friendly, the machinery behind it is often fraught with challenges.
[00:02:35] First, there is the persistent issue of machine translation quality.
[00:02:39] While Genai has improved the standard, the output suffers from hallucinations, invented content that deviates from the source in creative writing. This might be an interesting feature in our industry where accuracy and brand trust are paramount. It is a non negligible risk we cannot afford for a wellness app button or an offer disclaimer to be translated creatively rather than accurately.
[00:03:06] In parallel, we had a deep hesitation on ethical grounds. The current wave of generic LLMs suffers from what we might call an ethical supply chain problem.
[00:03:17] Many public models are built on questionable content scraping technologies and the mass ingestion of translation memories without consent to train the models. This new technology consumes the intellectual property of human professionals, creators and translators who are rarely if ever remunerated or even credited.
[00:03:39] So we asked ourselves, is it a healthy choice to build our work efficiencies on the foundations of appropriated labor? Is it ethically sustainable to use tools that take advantage of the professionals who make global communication possible?
[00:03:54] Finally, there is the environmental cost.
[00:03:57] AI data centers demand an enormous amount of energy that in many cases does not come from renewable sources beyond electricity. These facilities also require a staggering amount of water for cooling, often competing with local communities for fresh water in drought prone areas.
[00:04:16] In the context of the global climate emergency, using resources at an unsustainable rate threatens the ecosystem that supports us all.
[00:04:26] As a localization team in a company dedicated to well being, we could not in good conscience take a road that would lead to savings and efficiencies at any cost.
[00:04:36] Redefining AI Intelligent Automation with all this in mind, we went back to the drawing board. The directive was to use AI to increase efficiency, but the specific tool was not prescribed. We asked a pivotal what if AI didn't mean Gen AI in the popular sense?
[00:04:57] What if we interpreted the mandate as intelligent automation? Instead, we decided to look inward.
[00:05:03] We realized that our concerns about using AI public models didn't mean rejecting automation entirely, and we weren't starting from zero.
[00:05:12] Over the years, our in house localization team built a massive repository of high quality tms.
[00:05:19] Much like building muscle memory through consistent discipline training. These assets were created and maintained by our human experts through years of hard work.
[00:05:30] We also had a robust technological stack based on phrase, a platform we had already tailored to our needs and also integrated into our ecosystem. The adoption of this technology and our committed work had already generated production efficiencies.
[00:05:46] But if we were to take things to the next level we while maintaining our language operations well being, we would need to consider using technology in a smart, innovative way. This exploration led us to our solution phrase's custom AI.
[00:06:01] This choice represented a fundamental divergence from the Gen AI for everything trend. Unlike generic models that act as black boxes trained on the entire Internet, this technology trains neural machine translation NMT engines exclusively on our own linguistic assets. This option offered several critical advantages.
[00:06:24] First, by training the model strictly on our own curated tms, we eliminated the risk of Gen AI hallucinations since our technology of choice relies entirely on our approved terminology and brand voice, improving reliable, highly accurate outputs and reducing creative deviations.
[00:06:44] Second, by following A walled garden approach. We could select the materials used to train the model, ensuring that we were controlling the process and also that our data never leaked into public models.
[00:06:57] Likewise, using RTMs meant we would not have to rely on scraped content from the Internet. That would allow us to avoid normative risks regarding ip, among others.
[00:07:09] Finally, the approach achieved improved environmental sustainability.
[00:07:13] The energy and water demands for smaller scope NMT models are considerably lower than gen AI models.
[00:07:21] Holistic Implementation Choosing the right approach to technology was only the first step to make intelligent automation work. We had to ensure our inputs were pristine.
[00:07:33] As the old adage goes, garbage in, garbage out, we knew that an AI engine is only as good quality as the training data, so we followed a strict protocol to ensure our automation remained high quality and reliable. The process began with a comprehensive data audit. RTMs are the result of relentless and meticulous work over years, but like any legacy database, they weren't perfect. Before we trained a single engine, we looked inward and performed a rigorous data detox of our existing assets. We removed outdated terminology, fixed historical inconsistencies, and aligned our segmented data with our current brand voice.
[00:08:17] By ensuring that the training data was pristine, we guaranteed that our technology would learn from the best version of our work rather than replicating our past mistakes.
[00:08:27] Once the data was clean, we focused on glossary integration.
[00:08:31] Wahub's terminology is highly specific to the corporate wellness space.
[00:08:36] For instance, eligibles is the term we use for the employees who have access to our platform.
[00:08:43] Likewise, members is not quite the same as users. We invested heavily in updating our glossaries within our platform and ensuring they were consistently enforced by the technology engineering. This step was crucial to prevent the translation of words based on probability.
[00:09:00] Instead, it adheres to our specific corporate lexicon, ensuring consistency and quality across all user touchpoints. Finally, we designed the workflow to support adaptive learning.
[00:09:13] The true power of this NMT technology lies in its ability to improve over time.
[00:09:19] The system is not static. When our internal linguists edit AI output and make a correction, that correction doesn't just fix the sentence at hand, it feeds back into a TM that will be used again in the future to retrain the model periodically.
[00:09:35] This means that if we change a preferred turn today, the technology will learn about it and implement it in its future output. This creates a virtuous cycle where the more we use the system, the the smarter it gets and the less our humans have to correct the same errors over again.
[00:09:52] The proof of concept theory is necessary, but not enough.
[00:09:57] We also needed to prove this worked in practice.
[00:10:01] After some small controlled tests with promising results.
[00:10:05] We skipped a small scale pilot, instead applying the new workflow directly to the State of Work Life Wellness Report, one of our most important marketing efforts. Targeting executives and human resources leaders, this annual 20,000 word publication offers thought leadership on understanding, measuring and improving employee well being and boosting worker satisfaction, retention and performance. Along with it, it was a great opportunity to determine whether our new approach could handle volume and complexity without sacrificing the nuanced quality required for such a high profile publication.
[00:10:43] In this new workflow, instead of outsourcing the report to an external vendor for translation, we processed it with our AI technology.
[00:10:52] Then the NMT output was edited by our team of lead linguists.
[00:10:57] The result?
[00:10:58] A resounding success.
[00:11:01] By removing the dependency on language service providers and using our custom engine for the first pass, we dramatically accelerated our timeline, reducing our complete turnaround time, external translation, and in house review by almost 50%.
[00:11:18] It's important to note that this speed did not come at the cost of lower quality. Because the engine was trained specifically on our data, the raw output already kept our usual high quality standards, requiring significantly less remediation than generic engine output would have.
[00:11:36] Apart from the significant spend reduction associated with the automation, our efficiency gains were transformative. Overall, even though editing NMT output takes slightly more time than human translated content, we not only improved production speed while maintaining quality, but also eliminated any administrative time invested in vendor management, requesting quotes, emailing files, chasing deadlines, feedback rounds, etc.
[00:12:06] This allowed us to save additional bandwidth, shifting those resources toward value generation.
[00:12:12] As a result, our workflow pivoted from a transactional model of outsourcing to an internalized model of expert review.
[00:12:21] Elevating Human Expertise these days, there seems to be a shared idea about our industry that technology has rendered human expertise obsolete, that the role of people is to simply polish any content provided by the machine.
[00:12:37] Our experience proved the opposite. Technology not only did not replace our localization team, but elevated it, making it more relevant to our company operations than ever before.
[00:12:49] Under our previous hybrid model, our internal localization leads primarily functioned as quality gatekeepers. They were tasked with editing and polishing translations delivered by external vendors, which often meant we were paying twice once for the vendor's translation and again for our internal lead's time to review and manage it.
[00:13:11] Now our localization experts possess full ownership of the content localization pipeline.
[00:13:17] Our technology handles the heavy lifting of the first draft, reducing the cognitive fatigue of repetitive manual corrections.
[00:13:25] This terminology compliant output helps our team flex their creative muscles and focus entirely on the last mile of quality and stylistic fluency, a task requiring true human nuance, empathy, creativity, and cultural understanding. This shift redefined the daily roles of our localization team.
[00:13:46] Our lead linguists have more time and capacity to ensure our core mission is intrinsic to every piece of content we publish. They act as cultural sentinels, catching idioms or concepts that might be technically correct but culturally jarring for a specific market target audience, asset type, or comm channel. Furthermore, they're the trainers of our own engine. They clearly see that their feedback directly improves the system, giving them a sense of ownership over technology versus the prospect of being replaced by it. We proved that keeping the human in the localization loop isn't just an ethical safeguard it is a quality necessity.
[00:14:27] The Dividend of Ethics the ultimate test of any strategic pivot is the bottom line.
[00:14:34] For us, our savings came from automation efficiencies and technological sovereignty. We achieved the company's goal of doing more with less, without compromising on our quality standards or our ethics.
[00:14:47] The financial impact has been immediate and measurable.
[00:14:51] In 2025 alone, during the first phase of our intelligent automation rollout, we achieved operational savings of approximately 40% of our yearly budget. As we migrate more content types to this workflow and the engines mature, we project these savings to increase, reaching an estimated 70% in 2026. But the return on investment goes beyond just the dollar figure.
[00:15:17] Apart from cost savings and eliminating our dependency from external vendors, another key outcome is that our time efficiencies are converted into extended bandwidth more quality focus hours that are invested into reviewing more content.
[00:15:33] This means that we have achieved a massive increase in operational, both time and costs efficiency by breaking the link between scope and budget. Under the old model, expanding our scope meant a linear increase in costs.
[00:15:48] If we wanted to do more, we had to pay more. Today, we have managed to move away from that pattern. The time efficiencies of our technology augmented workflow means we can significantly increase our scope, translating more Help center articles, more blog posts, and more user communications. And the zero cost investment eliminates a corresponding spike in costs.
[00:16:14] We have transformed the localization team from a cost center into a strategic partner that enables the business to move fast by being smart.
[00:16:23] Wellness in all Respects the conversation around AI in the language industry is often polarized. Parties are often categorized as either Luddites resisting inevitable change, or tech evangelists willing to automate everything at any cost.
[00:16:40] However, our journey shows there is a third path. We prove that you don't need to scrape the entire Internet, appropriate unpaid labor or risk your data in a public black box to get results by using our own data, the fruit of our own hard work to train specialized engines, we built a language automation system that is secure, compliant and highly effective.
[00:17:05] An important qualifier, intelligent automation is not a plug and play solution, it is a dividend paid on previous investments. Our ability to train an effective AI engine was based on years of clean tms produced and curated by human experts for companies starting from scratch. Without this repository, the cold start problem is real. We suggest a middle path.
[00:17:29] Start by working with freelance professionals to build your initial linguistic assets with a long term vision of eventual hybrid reviews rather than seeking immediate shortcuts. Ultimately, our localization journey mirrors the holistic mission we promote. The MKTL10N team realized early on that we would not be building a true global well being platform if we were using tools that inherently cause strain through intellectual exploitation or environmental degradation.
[00:17:58] Just as we advise our users to nourish their bodies with healthy habits, we chose to nourish our custom AI engines with meticulously curated linguistic data rather than scraping the digital data equivalent of junk food.
[00:18:13] Similarly, just as we advocate for mental health and work life balance, we deliberately designed a workflow that eliminates tedious burnout inducing busywork. By giving our linguists ownership over the engine's training, we took away the industry wide anxiety of AI driven job loss and replaced it with genuine empowerment.
[00:18:34] We proved that true operational efficiency isn't about ruthlessly cutting costs at the expense of the people involved, but rather about cultivating a healthy ecosystem where both technology and human talent thrive together. When you put well being at the core of your operations, doing more with less doesn't mean doing it with less care, it means doing it with more purpose.
[00:18:57] This article was written by Josevi Abad. He is an advertising translator and transcreation specialist and the former Director of Language Services for the Americas at global production agency Hogarth Worldwide. He currently leads localization operations for the marketing division at WellHub as the director of Localization and Global Experience Originally published in Multilingual Magazine issue 251April 2026.
