Some projects push your technical skills. Others remind you why you got into this work in the first place. Our engagement with Tabiya and Fundación Empujar did both.
Tabiya is a non-profit born out of research at the University of Oxford, building open-source AI tools that help young job seekers discover their skills and connect with economic opportunity. Their flagship product, Compass, is a conversational AI that does something deceptively simple: it talks to you about your life experience and translates it into a professional skills profile, including skills gained through informal work, caregiving, and self-employment that traditional systems ignore entirely.
Backed by the World Bank and selected for the Google.org Generative AI Accelerator, Compass had already proven its value in Kenya and South Africa. The next step was Latin America, starting with Argentina through a partnership with Fundación Empujar, a foundation that has helped over 6,000 young people enter the workforce since 2013.
That's where Besolvit came in.
What We Built
Compass was an English-only research platform when we started. Our job was to make it operational for the Argentine market, and that meant far more than translation.
Multilingual AI. We added full multilingual capability to the platform so conversations flow naturally in Spanish. This wasn't a localization layer on top of the existing system. It required reworking the AI pipeline to handle Spanish-language skill extraction, taxonomy mapping, and conversational flows from the ground up.
Faster, leaner conversations. The original Compass experience was designed for a research context. We re-engineered the conversational AI to be shorter and more efficient, optimized for mobile-first users. This was especially important given the target audience: young people between 18 and 30 who won't sit through a long, academic-style conversation. The experience needed to feel quick, relevant, and respectful of their attention span, or they'd simply drop off.
Admin tools for scale. Fundación Empujar needed to manage large groups of users, not just individuals. We built custom admin features including bulk report downloads, turning a research prototype into a tool their team could run day to day.
Measurement infrastructure. You can't improve what you can't measure. We integrated Google Analytics and Tag Manager to track every meaningful event in the user journey, from first message to completed skills profile, giving both Tabiya and Fundación Empujar the data they need to iterate and prove impact.
What's Ahead
Fundación Empujar is preparing to launch the first cohort of 4,000 users in March 2026. For many of these young people, it will be the first time an AI tool recognizes the skills they've built outside traditional employment and turns those skills into a path forward.
This project sits at the intersection of everything we care about at Besolvit: open-source technology, meaningful AI, and products that work for the people who need them most. We're proud to be part of it.
Why It Matters
The global conversation around AI tends to focus on productivity tools for knowledge workers. But some of the most impactful applications are the ones that reach people who have been invisible to technology until now: young job seekers in emerging markets, people with informal work histories, first-generation professionals.
Building for that audience requires a different kind of engineering. It demands empathy for the user, respect for the domain, and the discipline to make complex systems feel simple. That's the work we do at Besolvit, and this project is a case study in why it matters.