Learning new languages is challenging for humans. Whether learners use a language application or receive structured lessons from teachers, acquiring new languages requires hours of rote memorization and regular repetition.
This tedious and impersonal experience was the genesis of Norby, an AI companion robot fueled by large language models (LLMs), the generative AI engines propelling 80% of organizations’ business strategies, according to Dell Technologies research.
Norby, who at 10.6 inches tall is small enough to perch atop a desk and provides a personal touch to assist with language development. Today Norby sports more than 50 text-to-speech voices, offering a customizable experience that helps it appeal to a broad swath of language learners.
How rote memorization inspired Norby
For the record, Norby prefers the descriptor, “conversational language companion,” whose origin story comes from a deeply personal place.
Norby Founder and CEO Adrian Mullan, whose mother is Chinese, grew up learning how to speak Mandarin—a challenging language to learn. Some 30 years later, Mullan was reminded of his own dislike of rote learning when he began to teach his own daughter Mandarin.
Believing that a more engaging experience would prove more effective for new language learners, or those requiring speech therapy, Mullan in 2016 decided to build an interactive AI robot that teaches individuals based on their needs and preferences.
This proved to be no mean feat.
For Norby to converse naturally, it needed to be able to hold a conversation based on the user’s interests and desired outcome. Whether Norby was working with an eight-year-old boy in Tokyo learning English, or an 80-year-old stroke patient learning to speak again, it would need to support different accents and levels of proficiency, as well as comprehend local slang and idioms.
The work appears to have paid off, as these demo videos can attest.
“We have the ability now with some of the AI technology to tailor the interaction to each individual user,” Mullan said.
The tech behind the talking AI robot
Powering Norby requires more than just a single AI model. At its core, Norby combines multiple large language models with a custom software layer that handles memory, real-time speech recognition and adaptive personalization. This allows conversations to feel natural, fast and responsive to each user’s goals.
On the hardware side, Norby integrates an array of high-quality microphones and speakers inside its compact, barrel-shaped body, optimized for clear voice input and expressive output. Dedicated compute on board ensures that much of the processing happens locally, reducing latency and keeping interactions smooth—wrapped in a design that feels approachable and playful.
Navigating these complex processes requires robust partnerships. Working with Dell Technologies and NVIDIA, Norby elevated three distinct areas of its business: building the intelligent learning function, designing the physical robot device and producing marketing collateral.
To develop the learning function, data scientists used Dell Pro Max workstations equipped with NVIDIA RTX™ GPUs to run and fine tune multiple LLMs locally. These capabilities enabled Norby to customize human-to-computer conversational interaction, as well as support multiple languages, personalized user responses and real-time feedback for continuous improvement.
“With AI technology on Dell Pro Max workstations, we can tailor how Norby interacts with each user,” Mullan said. “ This makes it more engaging to learn or relearn a language.”
The Dell Pro Max workstations helped Norby expand from supporting 20 languages to having the capability to teach more than 40, with further languages to be added regularly.
Meanwhile Norby’s product designers used the Pro Max workstations and CAD applications to create precise 2D and 3D models. And for marketing, Norby leaned on the NVIDIA RTX™ GPUs to accelerate animated renderings in short videos displaying Norby’s different components.
Underpinning these solutions is the the Dell AI Factory with NVIDIA, which brings together Dell AI infrastructure and NVIDIA’s GPUs, networking and software with automated workflows to help organizations securely develop and deploy generative AI at scale.
Although Norby is currently in the launch phase, the company is already looking to expand from pilot programs in language learning centers and elderly care facilities to schools and hospitals. Eventually, Norby will offer a version of its AI companion that serves as an interactive information kiosk for companies, as well as a concierge in hotels.
Features and functionality aside, the speed bump Dell and NVIDIA’s AI-optimized technologies provide offer Norby a major advantage. “The speed increases we’ve seen from AI over the last 18 months have been dramatic,” Mullan said.
The bottom line
AI workloads are as varied as the environments they run in. Flexible, fit-for-purpose infrastructure serves as the foundation upon which AI can be safely and cost-effectively built.
No matter what your environment looks like, Dell and NVIDIA can help your organization build and scale AI systems. The Dell AI Factory with NVIDIA provides a consultative approach to deploying AI, paving the way for better business outcomes now and in the future.
If the Dell AI Factory with NVIDIA works this well for a startup like Norby, just imagine what it can do for your organization.
Learn more about the Dell AI Factory with NVIDIA.
