He spent a quarter of a million dollars trying to teach his employees English. It didn't work. So he built the thing that would.
Diego Villegas runs Slang from Cambridge, Massachusetts, where the company teaches a peculiar and badly underserved subject: the English you need to do your job. Not the English of ordering coffee or asking for directions, but the English of an accounting close, a sales call, a healthcare handoff, an HR review. Slang is an MIT-born platform that uses machine learning and natural language processing to build industry-specific English courses - more than 150 of them - for companies like Nestle, Banco Santander and Hyatt across more than a dozen countries.
The reason he cares about this so much is that he once paid dearly for the gap. Before Slang, Villegas spent over $250,000 on English programs for the team at his previous company and never got the results he was after. To him that wasn't a budget line. It was evidence that an entire market was broken.
Most founders describe their origin story as a clever insight. Villegas describes his as a frustration he couldn't shake - and a phrase he kept coming back to.
“Man, this is professional illiteracy.
Diego Villegas, on the problem Slang exists to solve
It is a blunt way to name something polite companies tend to tiptoe around. Plenty of capable people can chat in English and still drown the moment the vocabulary turns technical. A brilliant engineer who can't write a clear status report. A sharp analyst who freezes on an English-language conference call. Conversational fluency and professional fluency are different skills, and Villegas built a company on the difference.
Figures drawn from investor write-ups and press coverage of Slang's 2021 Series A round.
In 1993, the same year he finished his undergraduate studies at Universidad de Los Andes, Villegas founded MASA - a technical-services and asset-integrity firm serving the Oil & Gas, Chemical and Power industries in Colombia. It was unglamorous, hands-on work: keeping the machinery of heavy industry running and safe. He kept at it for nearly two decades.
MASA grew into one of the region's leading technical-services providers, compounding at roughly 34% a year over 18 years and reaching about $200 million in revenue with more than 5,000 employees before Stork B.V. acquired it in 2007. Villegas stayed on through the integration as an executive and board member.
That is usually where a founder's story ends - with the exit, the title, the comfortable plateau. His didn't. In the post-acquisition shake-out, he watched capable employees lose their jobs largely because they couldn't operate in English. The lesson stuck harder than the windfall.
“The main problem for remote work is professional proficiency.
Diego Villegas, on the borderless workforce Slang is built for
In 2013 Villegas did something most successful operators never do: he went back to being a student. He completed an MBA through MIT Sloan's Fellows Program in Innovation and Global Leadership. On campus he met Kamran Khan, an MIT-trained computer scientist who had just won an iOS app development competition. Khan held a master's in Computer Science and Engineering focused on computer graphics and human-computer interfaces - exactly the technical other-half to Villegas's hard-won market problem.
Slang started not as a pitch deck but as an MIT research project: how do you use AI and NLP to make learning a language as efficient as humanly - and machine-ly - possible? The pair turned that research into a product, and Slang launched publicly in 2018. The bet was specificity. Instead of one generic English course, Slang generates coursework tuned to Engineering, Healthcare, HR, Management, Sales, Marketing and beyond.
The machine-learning engine isn't a marketing flourish - it's the unit economics. Khan has said that a course in accounting English once took a teacher more than 30 weeks to build by hand. With Slang's AI pipeline, the team can now stand up a comparable course in about a week. That compression is what lets a small company keep a sprawling, niche catalog current.
As a manager he poured more than $250,000 into English training that didn't take. Most people would chalk it up to a bad vendor. He read it as a market failure - and went to build the replacement himself.
Long before boardrooms, Villegas won first place in his country's Physics and Mathematics State Olympics three years running, from 1990 to 1992. The analytical streak shows up in a company that treats language as a data problem.
His first company sold technical services to oil rigs. His second teaches grammar to accountants. Different worlds, identical thesis: find the skill the workforce is missing, then supply it at scale.
Slang's technical DNA traces to a single moment on the MIT campus - Kamran Khan winning an app-development competition, and Villegas recognizing the engineer he'd been looking for.
In 2021 Slang closed a $14 million Series A led by DILA Capital and ALIVE, with participation from Salesforce Ventures, Roble Ventures, Impact Engine and the Inter-American Development Bank. The mix of commercial and development-finance money tracks with the company's dual identity: a real B2B SaaS business and a genuine economic-mobility play for the global workforce.
Bars are scaled for visual comparison, not to a single linear axis.
Slang's stated mission is to use technology to redefine language learning in the workforce - to put specialized English within reach of anyone whose career might otherwise be capped by it. For Villegas that isn't an abstract ideal. He watched it happen to people who worked for him, and he's spent his second act making sure the next version of that story has a different ending.
It is a tidy kind of symmetry. The founder who built a company around heavy industry now builds one around the soft, decisive skill that industry kept overlooking - the ability to be understood, precisely, at work, in English.