Every vendor claims to be ‘AI-enabled’, but what’s actually live versus what’s still on the roadmap? Fosway’s latest AI market assessment maps the real maturity of AI across learning systems, separating hype from reality. TrainingZone’s Becky Norman spoke to David Wilson, Fosway Group Research Director, to understand what L&D leaders need to know now.
Hype versus reality
Walk me through how Fosway’s AI market assessment came together. What prompted you to dig into this now, and what holes were you seeing in the information available to L&D teams?
We’ve been working on this for over a year, building on a long history of exploring AI as part of the broader innovation story in learning, HR and talent. This includes our Realities research, which examines the corporate view of AI, as well as extensive input from vendors through initiatives like the 9-Grid™ process.
What became clear was the sheer level of hype and noise surrounding AI. I sometimes describe it as a ‘tsunami of noise’. At trade shows, every vendor now positions themselves as ‘AI-enabled’ or ‘AI-first’. But our deeper Realities research revealed significant differences beneath the surface. ChatGPT’s rise at the end of 2022 accelerated this trend, and suddenly AI became a major part of every vendor’s roadmap.
The goal was to deconstruct this landscape and understand what was really happening, so we could provide practical guidance to corporate L&D teams. We wanted to help organisations assess the real opportunities, understand their current position, and determine how far behind – or ahead – they might be.
Only 10% of AI features are actually live
Your research shows that only 10% of mapped AI features in learning systems are actually live. Are vendors over-promising and under-delivering, or is something else at play?
It’s a reflection of the varying levels of AI maturity among vendors. Some have been investing in AI for many years, and a few are genuinely AI-native. However, the vast majority are not – they’ve just added AI to their roadmaps. What we’re seeing is a gap between the volume of noise and the reality of delivery.
There are many different features and capabilities that AI can impact, and vendors often have ambitious aspirations. But the reality of implementing those ambitions means that relatively little is actually live with customers today. In learning systems, the level of noise is very high, but this helps us gauge the true maturity of vendors and their ability to deliver.
The proportion of genuinely mature AI capabilities remains relatively small. We expect that to change over time.
Which specific claims or promises are you most sceptical about right now?
The biggest reality point is that much of what’s being done isn’t particularly transformational. Most mainstream AI features simply replicate existing processes, using AI to automate or enhance parts of them, rather than introducing entirely new ways of working.
In learning systems, the only mainstream feature – live with more than 50% of vendors – is smart recommendations of content for learners. That relates to contextually specific suggestions typically powered by machine learning rather than by large language models like ChatGPT. This capability has been in many systems or in development for a long time.
What I’m most sceptical about is the contrast between the story around AI, which is often framed as transformational, and the reality, which is actually quite transactional. In practice, AI is being used to accelerate or optimise existing processes, rather than rethinking them.
Learning experience and management: The ‘hot zones’
Learning experience and learning management – that’s where the money and attention are flowing. What makes these two ‘hot zones’ so attractive to vendors?
Learning experience as the front-end element has historically driven much of the AI investment, particularly in machine learning. The one mainstream AI feature across all areas is smart content recommendations – suggesting relevant content to learners based on their job roles, skills, learning activities and behaviours. It has been part of the learning experience story for a long time. Learning experience also has the highest proportion of live AI features with customers.
Learning management covers a wide range of areas: formal learning processes, training management, classroom delivery, integrations and more. Its higher AI presence is largely due to processes related to learning content, such as content creation, translation and management.
It’s less about learning management as a whole and more about learning content within it. Other areas of operational learning management are still largely untouched by AI.
If I’m an L&D leader wanting results this year, not three years from now, which AI capabilities should I actually be looking for?
That’s quite contextual to each organisation and its specific challenges. Typically, the leading topics we see are related to skills – upskilling, reskilling and understanding skills. This is a major priority from an AI perspective but also a complex one; it’s not an instant fix.
Learning experience is another key driver. AI is having a real impact here, particularly through emerging AI-enabled coaching interfaces. We’re starting to see a shift from the traditional ‘Netflix’ learner experience toward AI as a primary interface – helping people discover relevant learning opportunities and assemble tailored answers for them.
Another major area is learning content. AI is increasingly used to accelerate content production, translation, media and asset creation. There’s been significant uptake in this space, and it’s enabling L&D teams to quickly adapt their operating models – reducing reliance on external providers and speeding up content development from internal SMEs.
These are the most immediate areas of impact. Skills clearly has huge long-term potential, but getting skills right is challenging; it typically takes more than a year to get to a point where you’ve actually got a useful skills strategy.
The ‘cold zones’ – and why they matter
Tell us about the ‘cold zones’ – where AI progress is less advanced. Why are these areas being neglected for now?
The obvious areas are things like operational services and back-end support, or the administration and management of learning as a strategic capability, rather than at the course or content level. Additionally, developmental learning has huge potential – AI could play a significant role in supporting learners beyond basic knowledge or skill acquisition.
These areas are currently very underplayed. Why? Probably because other areas have been easier to tackle and implement. Only about 10% of learning systems have live AI innovation, which indicates relative immaturity. By comparison, talent acquisition systems are at around 30% live AI, so there’s a lot more maturity there.
These ‘cold zones’ exist because it’s not as easy to simply bolt on an LLM and generate immediate outcomes.
The edge advantage: What’s next
You’ve identified 28 features still on the ‘edge’, with less than a quarter of vendors offering them live. Pick one that represents a genuine opportunity for organisations willing to look beyond the mainstream.
First, it’s important to define ‘Edge Advantage’. We approached this fairly logically. We first identified what was mainstream – defined as live with more than 50% of vendors. Then what’s coming next, that’s ‘Next Wave’, where there is high consensus but less than 50% of vendors have it live.
Edge Advantage is the next segment. These are features with high vendor consensus (over 50%) but are live with less than 25% of vendors. This represents capabilities that most vendors want to offer, but few have live today. We called it Edge Advantage because these features are differentiating: a vendor that has several of these clearly stands out. And because there’s high consensus, they also have the potential to become mainstream in the future.
Some interesting examples include AI assistance in learning, such as virtual tutors, study buddies and adaptive learning. About two-thirds of vendors have adaptive learning on their roadmap; only roughly one in six has a live capability, and the effectiveness of that capability can still be debated. These are potentially transformational features that deliver significant value, but they are difficult to deliver effectively.
