26 October 2026

Your colleagues aren't waiting for permission: the AI opportunity for L&D

f X l @

Somewhere in your organisation someone is asking an AI how to handle a difficult conversation, explain a policy or pick up a new skill. They didn't wait for a course, a rollout or a login.

That isn't a threat to Learning and Development. It's a clear demand signal for what L&D does best. Two new studies on AI in L&D show it: the AI in L&D 2026 report from Donald H Taylor and Eglė Vinauskaitė finds over two thirds of L&D teams now using AI, up from around 40% two years ago, while OpenAI's usage data shows about a quarter of work-related messages on individual accounts are practical guidance, from how-to help to explanations and ideas.

People want help at the point of need – in the moments that matter. L&D can give it to them in a form that's secure, grounded in your own content and shaped by the people who know your organisation best. Here's what the research says, and how to make the most of the moment.

What the research says about AI and learning at work

L&D is building AI around its core work, while colleagues are looking for help at their personal point of need.

From the AI in L&D 2026 report (531 practitioners in 55 countries):

  • Content and learning design remain the most common uses, but evaluation and analytics has climbed from 8th to 3rd place since 2023.
  • The barriers L&D controls itself, such as skills and knowing where to start, fell from 49% to 36% in a year.
  • AI that reaches learners directly is at its lowest point in four years. Point-of-need support slipped from 7.7% to 6.2% of respondents.

And yet from OpenAI's usage data on individual accounts (July 2024 to June 2026):

  • About a quarter of work-related messages are practical guidance: how-to help, explanations and ideas.
  • UK usage has been more work-focused than the global average in every one of the last 24 months.
  • The share of UK messages from people aged 35 and over rose from 39% to 44% in a year, so this isn't only early-career staff.

It isn't only ChatGPT. Data published by other AI providers, mostly based on their consumer products, shows the same appetite for learning and support. Google's analysis of about 15 million Gemini interactions found that education makes up around one in five conversations in the Gemini app and AI Mode, and that information retrieval and learning is one of the main ways people use it for work. Anthropic's Economic Index shows explanations as the most common output in Claude chat conversations (17%), and 68% of surveyed users say they learn more with AI. Different tools, different users, same signal: people want to learn, and they want help the moment they need it.

This isn’t necessarily a story of L&D holding back. Respondents describe IT approvals, licensing and data privacy as the main obstacles to getting more AI tools supporting learning, and privacy is the most common barrier at 52%. Some L&D teams describe wanting to do more with AI and working through their organisation's rules to get there.

The opportunity: make the best option the safe one

People have shown they want help at the point of need. L&D can offer a version that is secure, relevant and managed by the people who know the content.

We've been here before. Surveys have long shown employees turning to Google and YouTube for help at work: a 2019 survey of 2,400 UK and US employees found more than a third did so rather than ask a colleague, and a recent Adobe survey reported about 36% of US workers going to YouTube for help learning agentic AI. People go where help is fast and specific. AI raises the stakes, because it generates answers rather than pointing to sources, which makes grounding them in your own content even more important.

That means AI that is:

  • Grounded in your content. Answers come from your policies, products and training, not the open internet.
  • Focused on a task or subject. A tutor or coach with a clear job gives more useful support than a general chatbot.
  • Managed and visible. L&D controls the sources and the behaviour, and can see how it's being used.

The aim isn't to replace the tools people already like. It's to give colleagues a trusted option alongside them, with human judgement still at the centre.

AI assistants, tutors and coaches in Breeio LMS

Breeio gives L&D teams a set of tools to put those three principles into practice.

  • Assistants trained on your files. Administrators upload and manage source documents, such as HR policies, IT security policies or product information, name the assistant and select the audience. Colleagues ask questions and get answers drawn from your content, for example what paternity leave they're eligible for and which dates matter.
  • Tutors and coaches with a specific job. Add AI tutors and scenario challenges to courses, each focused on a specific task or subject. They adapt to the course, the company, the industry and the colleague's role.
  • Coaching for development. A feedback coach supports reflection on practical modules, an objectives coach helps colleagues set and save objectives, and a learning plan assistant builds personal plans that update as people progress. A GROW-model AI Coach sits alongside traditional coaching, with access controlled by administrators.
  • Scenarios and content creation. Create decision-making scenarios that ask learners to explain their reasoning, and build modules and assessments from uploaded documents or a short brief, then edit them in Breeio's e-learning editor or Word.
  • AI usage reporting shows feature and credit use at a glance.

Security is built in. Breeio LMS uses Office 365 security and a private Microsoft Azure OpenAI model, and doesn't connect to the public ChatGPT. Your documents and answers are never shared with, or used to train, any AI model. Assistants can also be used in Microsoft Teams, where many colleagues already work.

How L&D teams can get started with AI

You can start small by asking four questions:

1. What are colleagues already asking AI for, and what do they wish they could ask your organisation?

2. Which content could ground an assistant, such as HR policies, IT security guidance or product information?

3. Where in your courses would a tutor or scenario challenge help most?

4. How will you see what's being used, and what feedback will you gather?

If you'd like to see Breeio LMS in action, book a demo or email hello@breeio.com. And we'd love to hear: what would your people ask an AI assistant if you gave them one?


Frequently asked questions

Are employees using AI to learn at work?

Usage data from OpenAI, Google and Anthropic suggests many are. Practical guidance, education and explanations feature heavily in each. The data comes mostly from consumer accounts, so it's a signal rather than a full picture.

What is an AI tutor or coach in an LMS?

It's an AI assistant given a specific task or subject to support a learner, such as guiding them through a course topic or coaching them through a scenario, using content the L&D team provides.

What is a performance support assistant?

It's an assistant trained on documents such as policies and product information, so colleagues can get answers when they need them. Answers come from your own files, not the open internet.

How can L&D make AI use safer?

Give people a sanctioned option that is grounded in your own content, managed by L&D and visible through usage reporting. Breeio runs on a private Azure OpenAI model, and your documents and answers are never used to train any AI model.

Sources