Stellar Labs and Faktion Accelerate Learning and Development Through Generative AI
Stellar Labs, an Antwerp-based start-up, enables organizations to deliver Learning & Development (L&D) journeys to their employees based on scientifically researched and proven neuroscience principles through an AI-powered SaaS platform. Recently, Stellar Labs partnered with Faktion to further expand its AI capabilities and begin building an MLOps platform – the original plan that we later had to pivot – to provide an even better experience for its users and deliver more business value by dramatically reducing the time spent creating journeys and providing them with personalized learning journeys.
On a mission to take people from knowing to doing
Stellar Labs is a software company that has built a robust Learning & Development (L&D) platform. Their platform serves as a comprehensive learning solution aimed at building and transforming skill sets in the workplace and driving tangible behavioural change in their workforce. The primary objective is to shift the paradigm from mere knowledge acquisition to the active application of acquired skills in the professional environment.
Traditionally, the L&D landscape has been saturated with content-rich platforms with little or no scientific backing, providing users with an overload of information and knowledge resources. However, they often fail to bridge the gap between training, application, and retention of newly acquired skills in the workplace, making training less effective.
In today's rapidly evolving business landscape, the need for upskilling and reskilling is constant. Organizations around the world recognize the need to continuously improve the skills of their workforce. However, traditional methods often fall short in terms of effectiveness. This is where Stellar Labs comes in.
How the stars aligned for a stellar collaboration
From the outset, it was clear that the two teams were highly complementary. The combination of Stellar Labs' vision on L&D and our ability to envision and productize AI applications, would create synergies. Stellar Labs has a strong team of software engineers, domain experts, and learning specialists in the L&D space. This team has know-how, particularly in the areas of software engineering, natural language processing (NLP) and data science (DS) in the context of L&D. In addition, Stellar Labs has a large number of L&D scientists with academic backgrounds and uses state-of-the-art learning techniques to be at the forefront in this domain. Our team at Faktion has extensive experience in shaping, applying and productizing AI, with a strong track record in NLP, LLMs as well as Machine Learning Operations (MLOps).
The combination of Stellar Labs' expertise with Faktion's capabilities in building MLOps platforms is a perfect fit. This synergy would unlock the next level of automation for L&D content. With this AI system, models can be trained with proprietary data and maintained by Stellar Labs, providing customers with more tailored learning journeys. All the plans, roadmaps, etc. were in place and ready to go, and then came the release of OpenAI's ChatGPT at the end of November and, more importantly, the release of GPT-4 in March.
It became clear that for many L&D applications, this new generation of LLMs and generative AI, accessible via APIs, could provide the content for parts of the learning journey. The implication for Stellar Labs however, at its stage of maturity, was that building an MLOps architecture was no longer a priority. Harnessing these LLMs and integrating them into their platform would be a huge leap forward and would give them significant tailwind. The project would be pivoted and the MLOps platform would be deprioritized - possibly to be revisited at a later stage of the project - and replaced by the integration of LLMs into the Stellar Labs learning environment.
Some of the key questions asked were: "How can we use this new generation of LLMs and respond quickly?", "How do we integrate it into our platform?", "How do we write the right prompts?”, “How do we continuously validate the output?”
Instead of developing the MLOps environment, the joint product team shifted gears and did conducted research on how to develop and leverage LLMs and in the context of L&D. Once the decision was made to pivot, the direction was clear: research, engineering a proof of concept (POC), scaling to a proof of value (POV), deployment using LLMOps, and further productization.
Under this new paradigm, and with the clock ticking in the arms race to adopt and leverage this new technology, the goal was to bring Stellar Labs' data science and LLM experience to the production level, and Faktion's NLP/LLM and productization experience to the specifics of L&D. Both teams worked together, building these capabilities in record time with unstoppable enthusiasm.
A New Position in the Market
The stars aligned for a stellar collaboration: a unique vision, complementary teams and the emergence of the new generation of LLMs that enabled Stellar Labs to become a first mover in the L&D segment. As the quality of the LLM output is key, continuous validation is paramount. It is not about generating more and more content, but about generating personalized content and learning journeys to train a specific set of skills. Therefore, continuous evaluation of the output is crucial, while keeping costs and complexity under control. In the past, the creation of learning journeys was a painstaking process carried out by Stellar Labs' L&D specialists over a period of weeks. Now the solution creates high-quality and detailed learning journeys much faster than before, at a fraction of the cost. The field of L&D has not changed as much in recent years. To be truly successful you need to generate learning, not just content.
As a first mover in productized LLMs for L&D, Stellar Labs is now taking its GEAR model (Guide, Experiment, Apply, and Retain) to stellar heights. Employers and users can rest assured that theory is actually being transferred and applied to further develop capabilities as well as learn new skills.
Conclusion
In conclusion, these LLMs are fundamentally changing the way up-skilling and re-skilling is done. Previously, personalization at an individual level would not have been feasible due to the significant investment required. Now companies can design a learning journey based on someone's current role, career goals and ambitions, as well as learning preferences. Some key take-aways:
- Never waste a good crisis. Pivoting in the middle of a project was a risky move, but a rewarding one with a significant payoff in the end. Being able to adapt and embrace disruptive technologies is the way to stay ahead of your competitors.
- When adopting a new technology, you should be aware of the potential as well as the pitfalls. In our case, LLMs are powerful but their output still have to be validated to avoid unreliable content. It is therefore important to constantly evaluate the models to ensure quality and relevant content and to keep them aligned with the Stellar Labs methodology.
- The fusion of domain expertise and technological innovation is key to creating and maintaining a competitive advantage. Cross-disciplinary collaboration results in a more holistic solution that is better able to address industry challenges. In addition, the parties involved can learn a lot from each other.
Stellar Labs and Faktion look forward to working together in the future, the stars are the limit.
"Faktion’s expertise and support was invaluable and perfectly timed. They have helped to integrate LLMs into our L&D platform, enabling personalised learning journeys and experiences for our users."
