In an era where language learning is increasingly intertwined with technology, Franz Ardito, Principal Product Manager at Babbel, shares the inspiring team journey into the development of our Babbel Speak tab, previously known as AI Conversation Partner, designed to ease the anxiety of speaking a new language.
With over 10 years of experience working on Babbel as a product, Franz has seen firsthand the evolution of language learning tools and the growing potential of AI.
In this interview, Franz discusses the inspiration behind our Babbel Speak tab and the challenges faced by the team during its development. Join us as we explore its story and its mission to help language learners gain confidence and fluency.
Let’s start with the basics: What is Babbel Speak?
Funny story – the concept of AI Conversation Partner, what eventually became the Babbel Speak, actually started back in 2019 as an internal Hackday project with the wonderfully chaotic name BabbelBlabberBlaster. The idea was bold: create a safe space where learners could practice open-ended, natural-feeling conversations without the pressure of speaking to another person.
We knew from both data and personal experience that one of the biggest blockers for new learners isn’t just a lack of vocabulary—it’s the anxiety that comes with speaking to native speakers. It’s an emotional barrier even before it’s a didactical one. The fear of making mistakes, of sounding foolish—it’s real. So we thought: What if we remove the barrier and give our learners a safe, private space to practice?
We wanted a space in our app that was available anytime, anywhere. We figured it was worth a try.
Cognitive science tells us that speaking is one of the last skills learners acquire when studying a new language, and it can be incredibly intimidating. With Babbel Speak, our goal is to help users visualize themselves actually speaking a foreign language. That self-image is powerful. The first few minutes with the app are absolutely key, because if you can show someone, right away, that they can do it, they’re far more likely to stick with it long-term.
Of course, back in 2019, the technology simply wasn’t there to support that vision. But that changed rapidly with the release of GPT-3, and especially GPT-3.5. Suddenly, we had the right tool to bring this vision to life.
From there, we focused on technical derisking, figuring out how to work with this powerful new tech, how to shape it for language learning, and how to ensure it was safe, supportive, and actually helpful. This product feature was born out of that journey: a conversation partner designed to help users practice real-life speaking, whenever and wherever they want, without fear.
What are some of the key features in the latest update? How do you think the new features will enhance the user experience?
With this update, we focused on giving Babbel Speak a much-needed visual and UX refresh, bringing it in line with the broader design evolution happening across the Babbel app. It’s more than just a new coat of paint: the feature has moved from an experimental “alpha” phase, where we were moving quickly, testing ideas, and validating value, to its current beta phase. This marks an important step toward a more polished, finalized experience.
Furthermore, we expanded to 200+ expert-curated scenarios across 8 categories covering real-life situations, made it prominently accessible through a dedicated home screen tab, and refined the pedagogy with clearer objectives and better feedback.
Ultimately, our goal is to create an environment that encourages regular practice and has the right scaffolding in place to support them. If learners enjoy how it looks and feels, they’re more likely to come back, and that’s where real progress happens.
Who is the team behind Babbel Speak?
Before I introduce, I want everyone to understand the background of this team: As mentioned before, the idea of creating a safe space to practice conversations has been with us for a while, during 2019’s Babbel Hackday. At the time, it was myself, Hector Hernandez (Computational Linguistics Engineer), and a couple of developers. It was a 12-hour sprint, but the idea stuck with me.
Fast-forward to 2022—when GPT-3 became available via API—Stefan Kainz (Software Engineer) and I picked the concept back up and started experimenting. We built a series of low-cost, low-fidelity prototypes just to understand what this technology could do, and more importantly, what it couldn’t.
2023 was a year of rapid iteration. We produced countless prototypes and proofs of concept, gradually converging on something that could be integrated into the Babbel app and tested with real users. That’s when Ioana, also known as Yoyo (Product Design) and Silvia (Content & Didactics) joined us, helping to shape the experience into something both pedagogically sound and delightful to use.
That core group evolved into what we jokingly call Team Asimov, named after the science fiction author whose work explored the relationship between humans and machines. I think Yoyo suggested that, quite fitting I think.
We operate like a focused innovation unit. Lean methodologies and a strong agile mindset, consolidated thanks to my now “partner in crime” Viktor (Engineering Manager) have guided us throughout the process: fast cycles, constant testing, and above all, prioritizing working software over lengthy documentation.
Today, our small original “Asimov” squad is blessed with the support of many other colleagues from product design (Ioanna & Ritu), engineering (Stefan, Mauricio & Akif), QA, content (Silvia), plus project management support from Shreya and, of course, our data team continue to shape and support Babbel Speak in a fully cross-functional setup. It’s been a collaborative journey and I think that energy is still very much alive: when things go wrong it’s the PM’s fault, when they go well it is a team effort – in our case it is truly a team effort.
What challenges did the team face during development, and how were they overcome?
A project like this naturally comes with a fair share of hurdles, starting with the unpredictability of working with emerging technology. From the very beginning, our challenge was twofold: making sure we were solving a real learner problem, and doing so with tech that was still evolving under our feet.
Limited resources, and the constant trade-off between speed and quality, were also daily realities. From my perspective as Principal Product Manager, one of the most critical things, especially early on, was setting up the process to generate learnings at every step. And ideally, generating those learnings as cheaply and quickly as possible.
Working with AI, especially LLMs, is quite different from traditional product development. I often compare it to working with water. You can guide it, but you can’t fully control or predict how it will behave. That’s why we deliberately took a small-batch, high-frequency approach: incremental steps, fast iterations, and constant testing to make sure the ground beneath us was stable before moving forward.
Sometimes this means optimizing for speed rather than accuracy, getting something scrappy out there, learning from it, and polishing later. The pace at which this technology evolves is insane. There are days when just staying on top of what’s new is a challenge, let alone prototype with it.
Our ability to respond to change, rather than just stick to a fixed roadmap, was a huge advantage. That flexibility is much easier to maintain in a small, nimble unit like Asimov. But scaling that mindset within a larger organisation, it’s a whole different challenge.
Looking ahead, what can users expect from Babbel Speak in future updates?
We’re staying in learning mode. Our focus is on making the beta experience smoother and more responsive based on real user feedback. The technology keeps evolving, and so will we: small iterations, constant improvements, always keeping learners at the center.
Is there anything else you’d like to share with our readers and learners?
I’ve been with Babbel for over 10 years now, and I still feel incredibly privileged to work on something so meaningful. As a product person, it’s rare to find a mission you truly believe in, and I believe deeply in Babbel’s mission: creating mutual understanding through language. Language is more than vocabulary and grammar; it’s connection. It’s empathy. It’s a bridge between people and cultures.
Learning a new language is also a deeply personal journey. It’s a powerful tool to project our own personality. When we learn our native language as kids, we’re shaped by our surroundings: family, culture, environment. But when we learn a new language later in life, especially as adults, we have more agency. We can choose the tone, the vocabulary, even the rhythm that best reflects who we are, or who we want to be. When I speak Italian, I’m a bit of a different Franz than when I speak English, Spanish, or German. It’s a rare opportunity to shape a version of ourselves in another language.
Working on Babbel Speak has been a way to bring that idea to life through technology. I’m a strong advocate of the Agile Manifesto and modern product principles, and I truly believe that if we stay grounded in those principles, move fast, learn constantly, and keep the user at the center, we can build something that not only works, but really makes a difference in people’s lives.
