Weeve is Building a Bridge Across the Language Chasm with AI-Powered EdTech
Weeve is Building a Bridge Across the Language Chasm with AI-Powered EdTech
EdTech founders Evan McGloughlin, Cian McNally, and Oisin Morrin are building a new paradigm for in-context language learning with artificial intelligence and neuroscience.
Are you learning a second language? Or a third? Language learning isn’t easy. “The most difficult hurdle in language learning is that gap between knowing nothing and immersion, and that gap is a chasm,” Weeve Languages CEO Evan McGloughlin told me just before the startup’s public launch this week. “Once you get to immersion, all you have to do is read and listen to content. 95 percent of learners don’t get that far.”
Founded in 2020 by childhood friends Evan McGloughlin and Cian McNally, Weeve is a Dublin, Ireland, based education technology company which aims to address the immersion gap. Today, Weeve launched its Chrome Extension, a tool which translates a portion of the words on any web page into the language students are learning.
Traditional methods seldom cater to beginners seeking comprehensive input, McGloughlin explained, depending on grammar exercises, memorization, or capsule lessons. Weeve relies instead on the diglot-weave technique, an approach which blends foreign words with one's native language in context. By offering a practical vehicle for beginners to read interesante content in context, it fosters the formation of a framework for a separate idioma.
Weeve has benefited from the revolution in artificial intelligence and large language models that has transformed society in the past year, overcoming challenges of scale and product-market fit while staying true to guiding principles of pedagogical integrity and community-led development, McGloughlin said. To date, the startup has won more than €250,000 in backing and commitments from leading Irish accelerator National Digital Research Centre (NDRC), several independent angel investors, and the High-Potential Start-Up (HPSU) program from Enterprise Ireland, a government agency supporting Irish entrepreneurship.
“We’re trying to build the best possible product for that chasm so that people can just immerse themselves and start living the language,” McGloughlin told Growing Season.
The Opportunity to Create the Paradigm
Cian McNally, the Chief Product Officer for Weeve, is fluent in four languages and self-describes as a “passionate polyglot.” He studied psychology and linguistic research at Trinity College, and first encountered the diglot weave technique while battling the steep language learning curve himself.
“Excited to put the method to test, Cian paid his little sister €20 to Google Translate Swedish words into Game of Thrones,” McGloughlin said. McNally quickly picked up Swedish, showcasing the promise of the method. “It was a good and fair test of language fluency, because if you don't know any Swedish and it works, then the method has some promise.”
McGloughlin, at the time an undergraduate in neuroscience, was convinced after reading The Hound of the Baskervilles with some words translated into Spanish. “I was fascinated by how well the brain grasps words in foreign contexts. Using our native language as a scaffold cleverly mirrors how we learn all new pieces of information, by weaving them into our existing mental framework,” he explained.
The possibility to scale a new educational paradigm to any language inspired both McGloughlin and McNally. Properly commercialized, the technique could revolutionize language learning. "I fell in love with that world – the neuroscience behind it,” McGloughlin said. The business potential was not lost on the entrepreneur, either.
We’re just getting people to watch Netflix.
“The language market is valued at €125bn, but learners are failed by insultingly inefficient techniques,” Weeve tells its investors. Finding an efficient, additive language learning method could hold immense value, but “weaving has not been commercialized before, I think, because it is actually technically very difficult,” McGloughlin told Growing Season.
The traditional and tedious approach taken by McNally’s sister – to manually and accurately translate selected words in a text tailored to a language learning level – are not scalable.
Even with the rapid developments in AI capabilities, “GPT-4's translation accuracy is a mere 50 percent, due to the lack of weave data online to inform the model. You can’t build an educational product on the back of that,” McGloughlin explained. Now, however, “you can create datasets tailored for specific language tasks, enabling the development of products according to your preferences.”
Weeve’s Chrome extension is based on proprietary AI technology, made possible by the recent revolution in large language models. McGloughlin and McNally see a browser add-on as an ideal medium for the diglot weave technique, as users would engage with the translated content whether Weeve was present or not.
“You don't need extra time to use Weeve, because it’s not like you’re adding another language-bearing tool like Duolingo to your belt. It’s just slotting into your life,” McGloughlin said. The company’s user statistics reflect that. “Our private beta users are engaging with Weeve for seven to eight hours a day now,” he said.
Language learning is a human endeavor. People need other people to learn languages fully.
As a polyglot, I was initially skeptical. My language mentors stressed understanding within the language for fluency rather than translation. Weaving seems unorthodox. I shared my doubt with McGloughlin.
“We're not saying Weaving is the key to fluency, but it's vital for crossing the initial vocabulary barrier,” he clarified. Weeve aims to help users acquire their first 3,000 to 4,000 words. “That is the most painful bit, and where we think we can have the most impact,” McGloughlin explained. Language follows a Pareto distribution, and “roughly 90 percent of spoken language is made up of the most frequently used several thousand of words.” Once you get to immersion, you can read, listen, and watch content in your second language. “We’re just getting people to watch Netflix,” he joked.
Additive pedagogy positions Weeve as a complementary tool to others in the edtech space, “We don't think that we're competing with Duolingo at the moment, which focuses more on grammar as opposed to vocabulary,” McGloughlin said, and competing is not the goal. Rather, it is creating a new paradigm for language learning and a new market for AI edtech.
“To have the most effective language journey, you need to be getting input from different sources so that your brain can process different things at once. It is a multivariate equation, and we want Weeve to be used as one tool in the arsenal.”
A Textbook Example of Finding Product-Market Fit
Weeve did not begin as a browser extension, or even an app. When McGloughlin and McNally first launched the business in 2020, it was under the name Diglot. “In the early days, we didn’t have the coding skills or money required to build the tech product we wanted to build,” McGloughlin told Growing Season. Initially, the pair translated out of copyright books using the diglot-weave technique – The Wizard of Oz, Pride and Prejudice, and Frankenstein among them.
The method was imperfect, but “Diglot was the most logical way to introduce our idea early and hear people's thoughts,” McGloughlin said. “It served as the entry portal into the Weeve vision.”
“The medium itself was imperfect for weaving because it could not be personalized or smoothly integrated into people's lives,” he explained. The sales volume across close to 2,000 customers reflected that, McGloughlin noted, as the company brought in “about 40 grand.”.
Struggling to find product market fit for what proved to be a good idea was an opportunity for testing and learning. “We suffered many terrible versions of Weeve to get to what we have now,” McGloughlin acknowledged. Candidly, a major blunder was “underestimating our own capabilities by not pivoting to digital sooner,” he said.
Instrumental to that pivot is Weeve’s now Chief Technology Officer, Oisin Morrin. “He's the most talented machine learning engineer in all of Ireland,” McGloughlin boasted. Morrin was at Trinity College when approached by Weeve to help create their translation algorithm. “We got some advice, and scoped the project for 700 hours,” he explained. “We were lucky enough to find Oisin. He built the entire thing in a weekend – over a Saturday and a Sunday. After he told us that, we said: "Okay, um, will you join us now please?”
Rebranding from Diglot to Weeve took place in October 2021. Led by UX & UI designer Aaron Connolly, the repositioning sparked a notable increase in customer retention. Following the branding push, Connolly joined the Weeve team full time.
“Previously, nobody could pronounce it,” McGloughlin joked. Over the next year, a pivot to community-led development increased customer retention, too. Weeve hired over 150 contractors around the globe to feed the language algorithm, ensure accuracy, and build a fully contextualized omnidirectional translation model.
A Community-Based Growth Strategy Acknowledges Language as a Human Endeavor
At the core of Weeve's nearly three-year-old identity lies a commitment to community building, McGloughlin emphasized. “Language learning is a human endeavor. People need other people to learn languages fully. At Weeve, we want to facilitate that human interaction.”
This philosophy comes to life through Weeve's product strategy, which centers direct user involvement. Though many startups foster community, emphasize product-led growth and consumer insights, and interact with their users directly on forums, McGloughlin insists Weeve goes deeper. Users are collaborators, not just passive consumers.
“Our suggestion board gathers input from our waitlisted users, fostering a sense of community while guiding product improvements based on user preferences,” McGloughlin explained. The Weeve community can track where requested features are in the development roadmap, and upvote or downvote others’ suggested builds. With limited development staff, listening attentively to users allows the Weeve team to prioritize what matters, creating enthusiasm and evangelism.
Recent additions like website blocks and grammar tabs exemplify the startup’s collaborative approach. The company’s efforts to weave the public into development encourage genuine connections beyond engagement with the app. “We're ready to go the extra mile,” McGloughlin emphasized.
Weeve’s growth strategy centers face time with the founders, too. They are the faces of the brand on social media, especially TikTok. “The results are impressive,” McGloughlin noted. “Over the past month, four of our videos went viral on TikTok.” Millions of likes and shares has translated into a near vertical lift in wait list sign ups ahead of Weeve’s public launch.
“Building personal founder-user relationships is something our competitors can’t steal,” McGloughlin emphasized. “It remains crucial to our overall moat strategy and defensibility in the coming years.”
A user-centered approach both solves for product-market fit, and generates evangelists. One user, a former business and law professor, now a renowned non-profit leader, runs a charity building schools for earthquake-affected children in Haiti. She was so impressed by the positive impact of Weeve on her French skills that she recorded a testimonial to be shared with Weeve's potential investors.
“Her video almost broke my heart,” McGloughlin said. “With the size of the challenge she has in Haiti, the fact that we can support her is incredible. It's stories like these that drive my passion as an edtech founder.”