Steal These Thoughts!

Is there anyone left on this planet who is not talking about Gen AI?

If there is, I’d love to have a conversation with them as I’m sure they’ll actually have something interesting to talk about from the usual discourse. Make no mistake, AI is interesting and a huge part of the present and future, but the bubble will burst.

Or it’ll just get bigger and The Matrix will happen in the next few years.

Saying all that…

Guess what we’re talking about today. Yes. Gen AI in the workplace.

What a hypocrite – forgive me! 🙏

I’m walking with you back in AI land because we (once again) are needed to unlock the mass adoption of AI at work. You read that right. 

Your company needs L&D to make this successful.

1. AI shapes tasks and jobs.
2. BCG’s report highlights the confidence and access gap in AI adoption
3. Focus on education, practical use cases, and equal learning opportunities for successful AI adoption at work
4. The 4-step plan helps build confidence, curiosity, and skills to bridge the AI adoption gap
5. AI usage yields time savings, increased speed, improved work quality, and more strategic work

---

Here’s what we know:

• Gen AI is era-defining tech just like the internet
• It’s reshaping how we do tasks and thus, how we build jobs
• Gen AI skills are a huge black hole for every company on earth

Yet, despite all the excitement. Not as many people are using AI powered tools as you think.

Why?

A new report from BCG sheds light on this.

Newsflash: People can’t use new stuff if you don’t show them.

We’re not in the business of magic.

As I see it, we have two problems to fix for mass adoption of AI tools at work:

• Improved education on the topic and practical use cases based on individual context
• Closing the gap between senior leaders and frontline employees access to knowledge

If we don’t do this then companies will literally be wasting millions, even billions, of dollars on the new wave of AI powered workplace tools.

Yes, my friend, it’s time for you to be the hero of this story.

Insights that matter from the BCG AI at Work Report

Ok. Let’s drop some knowledge to set the scene.

Those smart folks at BCG spoke with over 13,000 humans to find out what’s going on with AI at work today. As always, it’s packed full of stuff but these are the insights I want to focus on in our chat.

The adoption and confidence gap:

Turns out that the majority of us at the frontline where stuff gets done are not only less confident then our managers and leaders, we have less access to keep informed on new tools.

• Frontline employees: 33%
• Managers: 41%
• Leaders: 50%.

This becomes the 6th report I’ve seen findings like this.

I see it in my work with companies too. I shared more on this with a video on LinkedIn. We’ll get into strategies to fix this later.

The HUGE knowledge and skills disparity

Are you really surprised?

You probably shouldn’t be.

From what BCG tells us (much like this report from Asana). Those doing the work aren’t getting equal learning opportunities to craft skills with Gen AI. Leaders and managers are ahead of the humble frontline employee.

I find this odd as I believe those who spend most of the time doing the ‘actual work’ can benefit immensely from the new wave of AI tools.

The ‘expect people to figure it out strategy’ never works

It’s not lost on me that I’m preaching to the choir here.

It’s rare that people just get it. That’s where we (as in L&D and advocates of learning) come into play. Our role is not about skill building alone. We build confidence too. Confidence (and no BS knowledge) is exactly what the AI at work scene is missing.

Confidence builds curiosity, which leads to motivation which leads to learning.

From the BCG report, this is what we know:

• People don’t feel they have time to learn about AI (I was laughing so hard while writing that line)
• They aren’t sure how and when to use the tools that are available
• There’s a level of distrust with tools

4 steps to build Confidence, Curiosity and Skills to close the AI adoption gap

This is not going to be one of those useless lists.

Telling you to invest in continuous learning and be human centered is not going to help you. So, we’re going to get deep and tactical.

1/ Create an AI skills framework

To start, let me show you how I build a skills level framework for Gen AI. I use this with my clients to help them understand the meaningful learning experiences for each stage of maturity. Hopefully, it’s pretty self explanatory.

The goal is building confidence through meaningful knowledge and practical skills.

Skulls represent skill levels. You can deploy levels as singular projects, team-ups or a whole scaling journey. It all depends on workforce maturity and the context of the business.

2/ Discover where you are and where you want to be

To help me uncover the best fit for a business, I use three tools:

1. My AI skills readiness framework
2. The Asana AI maturity model (download avail below)
3. Discovery sessions

These are all free for you to use.

You’ll find detailed ‘how-to’ guides for each at the links provided above. These will help you understand the lay of the land today. With this information, you can make a recommendation on the best experience to deploy.

My experience as of writing this is that 99% are at the ‘New to AI’ stage.

That’s not a bad thing. It will enable you to put strong foundations in place by teaching the fundamentals and not jumping into whatever the current hot tool is.

→ Too many companies make that mistake.

3/ Create your strategy

You don’t need to overcomplicate this.

We’re not looking for perfection. Instead, you want a useful plan on a page to steer your AI ship.

Get some ideas on this in my detailed “How to build your AI learning strategy” guide.

Here’s a series of AI explainer videos from my YouTube channel you can use:

• How Generative AI Works: Everything you need to know about AI (7 part video series)
  
• How to work intelligently with Gen AI tools (11 part video series)

4/ Connect, share and learn

For true success, you need think outside the norm of L&D experiences.

Gen AI moves so damn fast that what we held as gold yesterday is worthless today. I get that can be scary. Its also an opportunity to encourage people to come together to cut through the noise.

My biggest recommendation is to build an internal community around this topic.

I don’t mean a company mandated one. No one wants to be part of that!

I’m positive you’ll find lots of curious and dubious humans who want to explore the Gen AI landscape.

Leverage tools like:

• Your company collaboration platform e.g Slack and Teams
• Encourage no-code with AI hackathons (you don’t need to be an engineer for this)
• Company emails and newsletters
• On-demand Gen AI toolkits with prompt templates and how-to videos.

What’s the ROI?

It’s easy for me to tell you to ‘do this’.

But it’s pointless unless it provides ROI for your work and business. The true monetary gains of Gen AI have yet to be realised. What we can see is the individual and team level benefits ↓

10,000 employees told BCG that using Gen AI has:

• Saved them time (84%)
• Increase speed to deliver (83%)
• Improved work quality (81%)
• Created time for more strategic work/human stuff (81%)
• Decreased time spent on admin (80%)

While this is impressive, I’m curious to learn what these people did with the time they got back.

Luckily, BCG covered that one ↓

 note they forgot to add ‘drink more tea’. I’ll let them off with that this time.

Get inspired: Case studies to erh…study

Everyone loves a good old case study, don’t they?

In case you missed it, I covered how 3 companies introduced AI tools to their teams. Specifically focusing on the problems they we’re trying to solve, how they deployed tools and tactics you can use too.

Final Thoughts

• Workforces won’t use new tools intelligently unless you show them
• Building confidence and dispelling fears enables people to build useful future skills
• Understanding the Matrix is pretty cool

Resources

---

Before you go… 👋

If you like my writing and think “Hey, I’d like to hear more of what this guy has to say” then you’re in luck.

You can join me every Tuesday morning for more tools, templates and insights for the modern L&D pro in my weekly newsletter.

It’s not your fault that you don’t know how to craft useful AI prompts to produce quality outputs.

We’re still very early in this game.

In reality, no one showed you how the language of these tools works so you don’t get what you expect.

I coined this the fail fast fallacy.

I also like the term prompt and ghost because that happens too much. It’s like people are in physical pain if they’re dared asked to ask another question after the tool response. Imagine if that’s how we talked to each other.

You gotta know how to talk to machines. A lot of people don’t and they blame the machines.

It’s human error – plain and simple.

If you treat a conversational AI like a google search your results will always suck – fact.

Fret not, we can help you if you’re reading this right now thinking “Shit, this is me!”. A little bit of education, guidance and practice will change that.

I’m going to break this down into these sections:

How we’ve navigated the digital world for 25(ish) years

Do you remember the time before Google?

I can. I was born in the late 80’s so my generation is probably one of, if not, the last to not come tumbling into this world surrounded by search engines, 24/7 connection and people posting dumb stuff for the world.

Some of you might be young enough to not know of this world.

While that makes me feel super old. You might find this nostalgic trip fun. I mean you missed the delights of AOL (America online) with it’s classic connection and browser.

Don’t get me started on the wonder that was “Ask Jeeves”. An early search engine moulded around a fictional english butler called Jeeves.

This was peak innovation and 12 year old me could not get enough of this virtual butler finding information for him. It could have been better at helping me amplify my Pokemon card obsession though.

Google enters the game

Everything changed in late 1998 when a startup called Google hit the scene with it’s ‘Google search engine’. It had been in testing behind closed doors at Stanford University since 1996 but late 98 is when us lucky everyday folk got hold of this era-defining tech.

You don’t need me to tell you the rest because you know it.

No one says ‘I’ll search the internet”, they say “I’ll Google that”. The ultimate category king.

The thing is Google has been our companion to navigate the digital highways ever since. We do everything from research, looking at cute cat pics, best restaurants in x city, looking up old friends to cry ourselves to sleep over how successful they’re and we’re not (def don’t do that) or to find that one YouTube video to fix your bathroom light.

Google has been the one destination we agree upon as “the place to do search”.

It’s remained the number one way to interact with the digital space, so by default, it is the number one language shared by each of us. That has worked well until now.

Before we get to that, let’s explore how this same tech lives in learning tech.

How this is replicated in popular LMS and LXP platforms

If keyword search has made Google a billion dollar company, why not use the same or similar tech for other content hosting systems?

That’s what a lot of tools have done for a long time now.

Most LMS platforms often use keyword search as their backbone content search infrastructure. I say ‘most’ because I don’t need you LMS providers who read this to be sending me hate mail – chill!.

Like Google, this help users find relevant courses and content. You enter your specific terms and the LMS will return results that contain those keywords in the course titles, descriptions, or content.

LXPs are a little different, or so they tell us.

Many use a combination of both keyword search, where search queries are unpacked into individual words and then matched within the content’s title, tags, and descriptions, plus natural language processing (NLP for short).

Natural language processing explained

This is not going to turn into the last part of the Matrix 2 where Neo and the architect meet to explain what the Matrix is.

I’ve watched that scene about 10 times and still don’t understand what the matrix is.

Anyway, back to NLP. This enables users to enter search queries in natural language, using conversational phrases or questions instead of just keywords. NLP algorithms process these queries to understand the user’s intent and context.

Dependent on how good the LXP is, a few things are available here.

Rather than just matching keywords, LXPs use semantic search which just means it can understand the meaning and context behind the query. This enables the platform to retrieve relevant content even if it doesn’t contain the exact words used in the search.

As an example:

Let’s say you search for “leadership skills for managers“.

With traditional keyword-based search, the LXP would look for content that contains those exact words or phrases within titles, descriptions, or tags. However, with semantic search powered by NLP, the LXP can understand the intent and meaning behind the query.

So, it recognises that the user is looking for learning resources related to developing leadership competencies for people in managerial roles.

The NLP algorithms would break down the query into its core components:

• Intent: Learn about leadership skills
• Target audience: Managers

It would then search the content library not just for literal keyword matches, but for semantically related concepts. This could surface results like:

• “Effective Communication for Leaders” course
• “Motivating and Inspiring Teams” video series
• “Emotional Intelligence in the Workplace” ebook
• Discussion thread: “What makes a great manager?”

Even if those content items don’t contain the exact phrase “leadership skills for managers”, the semantic search can identify them as highly relevant based on the contextual meaning extracted from the query.

Pretty smart.

Another useful feature is sentiment analysis which is deployed to understand the emotional tone or sentiment behind search queries, which can further refine the search results.

This is the same tech that google uses too. It’s like search on steroids. The thing is its still pretty basic knowledge retrieval which you as the human must assess.

The difference between a Google search term and AI prompts

For the uninitiated, when I say AI in this article I’m specifically talking about the generative family of AI models.

There are many different types of AI in our world.

Don’t let all the snazzy marketing campaigns fool you. To work with conversational Gen AI tools like ChatGPT, Microsoft Co-pilot, Google Gemini and the rest we must use a prompt.

A prompt is just an instruction.

You send an instruction to the tools database which then runs that query. No need to overcomplicate it here. Prompts need to provide a detailed, context-rich instruction or question to guide the AI in generating a specific response or performing a task. Someone classed this as prompt engineering.

There’s no actual engineering involved, fyi. Unless you class crafting words as engineering.

Ernst Hemingway might disagree.

The effectiveness of any conversational AI’s response heavily depends on how well the prompt is crafted. I cannot not overstate this. As of 2024, I’m sure this will be etched onto my grave.

A well-structured prompt will lead the AI through a step-by-step reasoning process, helping it understand the context better, and generate more accurate and relevant outputs.

If you want to get nerdier, you can deploy techniques like chaining prompts, where complex tasks are broken down into simpler prompts, and guided reasoning.

Like I said, it has a sort of language and art to it.

One that’s always evolving.

Google search terms

Using search terms on Google, on the other hand, involves entering keywords or phrases to find information that already exists on the internet.

The effectiveness of your search results depends on the choice of keywords and how well they match the content available online. So, if the marketer was smart enough to stuff their article with those key words, they win the search lottery.

Google’s algorithms analyse these keywords to retrieve and rank web pages based on relevance, quality, and other factors.

Unlike prompting an AI, where the goal is to generate new content or responses based on the input, using search terms on Google is about finding existing information.

⏱️ In sum: Prompting an AI and using search terms on Google serve different purposes and involve different processes. While AI prompting is about generating new content based on structured input, using search terms is about finding existing information through keyword matching.

How to unbundle yourself from the ‘google search mindset’ with AI

As L&D pros (I assume most of you reading this are), this bit should excite you.

To use conversational AI tools better, we must re-shape our mindset on what we’ve done these last 25 years. Easy, right? In essence, we need to learn new skills and behaviours.

Those are the things we’re in this game to do.

Before we sink our teeth into the final part of our chat, let’s look at steps you can take to unbundle your mindset:

1. Recognise how we’ve interacted with the digital world to date
2. Get clear on the difference between that and work with generative AI tools
3. Know when to use each method for what tool
4. Learn how to write AI prompts for any generative AI tool with the framework below

You’re welcome.

The universal AI prompt framework for critical tasks

Prompts for AI can come in many forms for different outcomes.

Looking for a quick answer to a simple question? Use a 1-2 sentence prompt. Want to unpack a complex work task with layers of actions? You need a different approach.

Whereas a one shot prompt like “What’s the weather in Ibiza on average in August?” is a simple string of words for any conversational AI tool to answer. Asking it to review, ideate and share how to create a learning strategy for your company is not so straightforward.

To get the best results from something like a ChatGPT, you need 3 things:

• Context
• Task outline
• Constraints

This is a universal approach you can use with any tool. Let’s unpack each of these:

Context

What does the LLM need to know to successfully support you?

Here’s some ideas:

• Your organisation
• Team
• Roles and work they do?
• Specifics on the task
• What have you done before?
• What is the role it’s playing? [If role-playing or coaching]

Task

• Outline the task
• What does success look like?
• What are the essential components of the task?
• Keep it clear and simple
• How should the output be structured? Bullets, sentences or paragraphs.

Constraints

• What should the LLM not focus on?
• What must it not consider?
• Should it only use its training data or connect to the internet or both?
• Should it only use the data you’ve provided?

If we put this all together, it can look something like this:

###Context###

I'm crafting my organisations L&D strategy for the year ahead. We’re a scale-up business with 800 employees in 5 global offices. We’re limited with our resource and budget to deliver. Our goal is not to do everything, but do the top 2-3 things that matter most.

Our strategy for the last few years has become stale and not working towards what we want to achieve.

Top things on our employees minds include:

- Having the right skills for the role

- Learning from their peers

- Manager support and coaching


###Task###

You will help me get clarity on how I can work with my team to build a relevant and meaningful strategy for our organisation.

To do this, you will ask me questions to help get clarity to build a better picture of where we can go.


###Constraints###

Keep questions short and relevant. They should be quick fire rather than in-depth. Let’s keep questions to a minimum of 3 at a time.

You might notice I’m using ‘markdown’ structure and delimiters in these prompts. This helps AI tools better understand the instructions you provide by using heading, bullets and general formatter structure etc.

You can learn more about that in this video.

3 prompting techniques you need to know

Let’s take some of the techy terms you might have heard and translate for humans.

1️⃣ Zero-shot

Useful for: Simple Q&A inputs

Example: “On average, how long do Alaskan Huskies live?“

2️⃣ One-shot

Useful for: When your task requires a specific output such as a table or executive summary with one examples as guidance.

Example: “I want an executive summary on the current population of Alaskan huskies in North America. The summary needs to follow our company format, here’s an example:

‘Insert your example’

Now, let’s create a first draft”

3️⃣ Few-shot

Useful for: Working on highly complex tasks like data analysis or writing a report for senior leaders using 2 – 3 examples.

Example: “I need to produce an analysis of the most popular buying locations of Alaskan Huskies in North America from European origins. This will need to be in our preferred company format with supporting data visualisations.

Here’s a few examples to show what I want to achieve:

*Either insert the examples directly into the input bar or upload as images, Docs and PDFs

Note: No prompt is bulletproof. LLMs can behave in extraordinary and odd ways. If you find yourself hitting that brick wall, deploy these techniques.

Final Thoughts

We’ve covered a lot of ground today.

• AI prompting is a new language you can learn
• Look at it as a universal meta skill for the next few years
• Use the write language/operating system for the right tech

If you’d like direct help from me on any of this, you can:

1. Take my prompt writing for business masterclass
2. Become the envy of your team in the AI For L&D Crash Course
3. Hire me to work with your team to craft an intelligent research based approach to AI at work

---

Before you go… 👋

If you like my writing and think “Hey, I’d like to hear more of what this guy has to say” then you’re in luck.

You can join me every Tuesday morning for more tools, templates and insights for the modern L&D pro in my weekly newsletter.

A big thing I notice in my work is the complete lack of understanding of how Generative AI technology works. Specifically, the science behind why AI makes up stuff.

For those who’ve taken my crash course, you know I focus on this as an essential block of knowledge.

The lack of understanding of why this happens and its unintended consequences leads to a lotta mistakes when applying tools at work.

Let’s fix that ↓

Why AI hallucinates

Not enough people understand that generative AI is a probabilistic system.

What does that mean?

Your favourite conversational AI tool is a probability engine.

Generative AI systems are prime examples of probabilistic models. When prompted, these models do not simply retrieve a pre-written response.

A probabilistic system uses data and patterns to make guesses about what might happen next.

It doesn’t always give the same result because it considers different possibilities and chooses one based on how likely it thinks that outcome is based on its training data.

How Generative AI creates answers

↳ Here’s how it works:

1. Learning from data: The model looks at a lot of examples (like reading many books or looking at many pictures) to understand patterns and relationships.
   
   
2. Making predictions: When you ask ChatGPT to do something, it uses what it learned to guess the best answer. It doesn’t follow a strict recipe. It considers different possibilities and picks one based on what it thinks is most likely to be correct.

For example, if you ask a generative AI tool to write a story, it will use patterns from all the stories it has read to create a new one.

The story might be different each time you ask because the AI is making educated guesses based on probabilities.

This probabilistic sampling allows generative AI to produce varied, open-ended outputs like paragraphs, images, or code.

Yet, it also means the outputs can be inconsistent or contain hallucinations since the model is essentially “guessing” the most probable output.

This is both a strength and a limitation of current generative AI technology.

→ It’s not a bug, it’s a feature.

Historically, search engines like Google have been deterministic systems. They use methods to find relevant information and give you the same results for that query.

You can’t use generative AI tools intelligently if you don’t understand this.

There’s a reason companies are worried about full-scale deployment of generative-powered assistants.

In sum: All these trendy AI tools have limitations. Be clear on these so you can use them intelligently.

Why this is important for L&D solutions and products ‘powered by AI’

Now I’ve pulled back the curtain on the matrix, you are better equipped to navigate how to use these tools in the L&D space.

Before you ask…no. It doesn’t matter how good your ’prompting’ skills are.

The capability for generative AI models to make up stuff is hard coded. As we’ve covered, it makes them both great and bad. If you go into using and building tools with AI with this at the top of your framework of ‘to use or not to use AI for this task’ you’ll do well.

A common example of mishaps can be seen with the thousands of chatbot solutions which have flooded the market.

Some of the worse offenders, imo, are the ‘coaching’ and specifically mental/health and wellbeing assistants. With what you know now:

1. Do you believe these areas are suitable for Gen AI support?
2. Are you comfortable with the 10% potential of bad advice?

Like any technology, Gen AI has it’s time and place for use.

Clarity on this = better solutions = improved performance

Final thoughts: TL;DR 📝

• What: Gen AI systems are probabilistic models that generate varied outputs based on learned data patterns and probabilities rather than fixed responses.
  
  
• Why: The ability to produce different results each time can lead to inconsistent outputs. Knowing this equips you to pick the right time to use Gen AI tools.
  
  
• How: You need to know these limitations to use AI intelligently.

---

Before you go… 👋

If you like my writing and think “Hey, I’d like to hear more of what this guy has to say” then you’re in luck.

You can join me every Tuesday morning for more tools, templates and insights for the modern L&D pro in my weekly newsletter.

This blog post isn’t just another list of the 25 million AI tools on the market today (99% are irrelevant and will fade away, FYI).

Instead, I want this to be more nuanced. I’ve refined this list to the 5 best tool for L&D teams. I’ve tested each extensively to ensure they’re worth your time.

You don’t, and certainly shouldn’t, use every tool on this list at the same time. Your job is to use the ones that work for your context.

In other words, don’t use ChatGPT, Claude, and Perplexity at the same time. Pick the one that works best for you through experimentation.

Sorry to those who were expecting one of those crazy visuals with millions of tools!

You don’t need more tools. You need the right tools, my friend.

Let’s unpack the best AI tools for L&D teams in 2025.

ChatGPT by OpenAI

You most likely have used or at least heard of this one.

Some might argue it is the best AI tool for L&D teams, but I won’t go that far.

It’s the poster child for the new wave of generative AI tools. I can’t deny its current top billing as the best-performing multi-model conversational AI tool.

It’s incredibly effective.

If Willy Wonka operated a delightful factory of AI tools rather than delicious chocolates, ChatGPT would be his most beloved creation.

What it can do:

• Generate new text
• Create an array of images
• Read different file formats including: XLX, PDF, TXT, PNG, JPEG, and more
• Create custom AI assistants that connect into other applications.
• Use voice and vision to talk with you and explore your surroundings

What you can use it for:

• A thought partner to support your ideas and general work
• Your research intern
• A data analyst that distills complex bloated reports into meaningful insights
• Cute cat images, if that’s your thing
• Live problem-solving with vision and voice mode to see the world around you, and share your screen.

My favourite use cases:

• Thought Partner
• LXD Assistant
• Research Assistant
• Data Analyst
• Skills coach

Perplexity

Perplexity is like the strategic digital L&D partner I always wanted.

Its support in data analytics and decision-making processes is unmatched. It simplifies and provides valuable insights without the need to write incredibly nuanced prompts.

What it can do:

• Conduct conversational search, generating answers from web sources with citations
• Summarise search results rather than listing them, prioritising newer sources to ensure up-to-date information
• Support file uploads for analysis, including text documents and CSV formats
• Refine searches to sources such as YouTube, Reddit, offline, and online
• Provide real-time data by surfing the web for the latest answers

What you can use it for:

• Problem solving with focused search features
• Browsing the internet for crafted answers rather than links (think Google 2.0)
• Uncovering unique insights and thinking critically with its follow-up deep dive questions

My favourite use cases:

• Research Assistant
• Intelligent data analyst
• Fact checker

NotebookLM by Google

This one is under the radar for most.

It’s a tool specifically built for research and analysis by the smart folk at Google. The USP of this tool is its ability to handle huge stores of data. It allows you to treat it like a notebook.

That means you can store little sticky notes of thoughts, discoveries and insights to revisit.

Key features include:

• Upload and analyse up to 300 different data sources in one chat
• Create FAQ’s, study guides and briefing docs from conversations
• Comparative data analysis – ask for critiques, breakdowns and related ideas on your research
• Generate podcast-style conversations based on your findings, voiced by AI-created hosts (weird, I know)
• Generate video overviews of your research

This is not your typical LLM.

It won’t create images for you, or any of the classic creative tasks the tools above are used for. That’s actually what makes it great. It does 1 – 2 tasks exceptionally well.

Learn how to use this tool with my in-depth tutorial playlist on YouTube ↓

ElevenLabs

The first of two non-specific conversational AI tools.

I wish I had access to ElevenLabs when I was leading a global team. It would have significantly improved the speed, quality, and cost of language translation.

I appreciate this tool because it supports access to content for all. Language is a big barrier when it comes to content.

What it can do:

• Convert text-based resources into audio
• Offer multilingual support for videos and audio
• Voiceovers for videos
• Dub your existing audio and video with your voice into any language

What you can use it for:

• Audio and video language translation at speed with quality
• High-quality voiceovers
• Dub your own voice into 29 languages

My favourite use cases:

• Dubbing my voice into any language – amazing yet scary at the same time
• Voiceovers for YouTube videos. I like to spice up my content.

Descript

A video editing suite so simple a 5-year-old can master it.

Of course, it comes with AI-powered features to simplify the creation and editing of visual content. It’s easy to use, saves time, and makes your videos look incredibly professional.

What more can you ask for?

What it can do:

• Edit videos with user-friendly, AI-assisted tools
• Auto–generate subtitles and transcripts for accessibility
• Repurpose existing video content into new formats
• Create podcasts and voiceovers with built-in recording features
• Clone your voice for speedy but slightly eerie creations

What you can use it for:

• Create slick videos with little experience
• Podcasts and audiograms
• Text-to-voice creations, if you’re interested in voice cloning

My favourite use cases:

• Producing all my video content – it’s my only video editing tool
• Adding subtitles in seconds to any video
• Adding studio sound quality to any video
• Editing my transcripts. This might sound simple, but it’s hard and time-consuming. Descript simplifies this.
• Generating b-roll, avatars and images with in-built AI models
• Using agent mode to help me design production elements I can’t do on my own

Bonus: HeyGen

Ok, I did put 5 and a bit in the title.

For a while, I had been resistant to AI avatar technology as I’d found its applications incredibly gimmicky, and most often used to elevate already bad L&D practices.

But the fact is that’s not the tech’s fault, it’s down to the users.

AI avatar tech has been getting better, and right now, HeyGen is the platform that most impresses me and where I get a lot of value when I do choose to utilise AI avatars.

It’s packed full of features, and they’re generally first to market with the latest innovations from the space. Check out the video below, where I create an interactive AI avatar that can listen and respond to users ↓

Build your AI toolkit

There you have it, the best AI Tools for L&D Teams in 2025.

The world of AI is an open playground.

So, experiment, explore, and do cool stuff.

If you feel like I missed one or you want to share your best AI tools for L&D teams in 2025 with me. Share this with me directly.

---

Before you go… 👋

If you like my writing and think “Hey, I’d like to hear more of what this guy has to say” then you’re in luck.

You can join me every Tuesday morning for more tools, templates and insights for the modern L&D pro in my weekly newsletter.

There’s a lot of noise on social with AI for L&D advice.

I find too much of it to be finite thinking, and not nearly enough on the enormity of the possible.

Too much focus on ‘perfect’ prompts (they don’t exist) and which LLM is the best. You could spend decades debating that one. It’s just like which LMS or LXP.

The answer is really simple – whatever one works best for your context.

Something that doesn’t get nearly enough attention is how those in the EdTech and LearnTech space can build useful solutions with AI.

I know, I know…there are tools. But a lot of them are pretty content focused, and we need more than that.

Let’s look at what we can build to enhance our work beyond content, and how we bring these to life. You’ll learn about the ideas, why they’re useful and how you can look to build similar solutions.

This is a playbook for you to get ahead of the curve.

Idea 1: Onboarding Assistant

Welcoming a new employee is an incredibly important event.

• Gallup found that a staggering 88% of employees surveyed said their org doesn’t provide a good onboarding experience.
• In their research, HBR found that one-third of employee attrition occurs within the first 90 days, with poor onboarding being a major factor for these departures. Additionally, they discovered that a further 23% of employees leave before their first anniversary for the same reason.

This is the first meaningful impression new employees get of the company. Sadly, not enough give it the attention it needs but it’s not their fault (in most cases).

Line managers are mostly underprepared, HR teams are snowed under with their daily adult babysitting and L&D is navigating a tsunami of orders across the business.

It’s hard, I know.

Years of designing, delivering and crying over onboarding experiences have given me tons of meaningful data.

Yet, what if we could enhance this process with the help of AI?

The problem we’re solving

Let’s be clear, I’m not talking about delegating the whole experience to AI.

That would be stupid.

I’m specifically looking at areas where we can provide quick in the moment support on low to mid-level tasks.

Stuff like:

• Learning more about the company
• Who to contact in xyz team
• How to complete x process?
• Where the f*!k are the printers and why can’t I connect to them?

Although these seem small, they’re a significant part of supporting newbies to become capable and confident.

They’re often the things that fall in the cracks because managers, HR and L&D don’t have time to attend to them all. With an intelligent approach with the latest tech, we can make sure these needs are covered.

The added benefit is that if these are covered, this gives managers and HR teams more time to focus on doing the human stuff with new starters.

One of my favourite AI apps to build for clients is an onboarding assistant.

I see a number of clear areas for conversational assistants to support performance and learning. The goal is not to replace human experiences, this is to enhance it.

In this video, I share why you might want to use a refined conversational AI assistant with access to company data to support onboarding and how to build the solution.

Idea 2: AI Agents handling pre-boarding backend tasks

This one is a little bit advanced, but I want to give you that ‘enormity of the possible’ view.

Note: You’ll need a tech team to help you build this type of solution.

After creating the video above, I started thinking about where else intelligent AI adoption could aid the A-Z onboarding process.

Then it hit me.

The problem we’re solving

A common problem I’ve seen in every organisation I’ve worked in is with the HR → Tech team communication for setting up new employees.

Things like, “Oh HR didn’t tell us this” or “IT didn’t send that software request form”. None of these things are helpful for a newbie to hear.

This is where a small team of AI agents (scroll down for an explainer on these) could help mitigate the space for human error to improve the experience for all teams.

This is an evolution of the onboarding AI assistant above.

Both can work together, but these serve different purposes.

By using AI to help with these types of tasks we can provide more time to focus on both employee and manager support.

At a high level (if you can’t be bothered to watch the video) this agent team looks like:

• A supervisor to ensure all tasks are completed
• An HR assistant agent to carry out HR tasks and communicate with the Tech support agent
• A Tech support worker to set up all equipment and software, and confirm everything is ready with HR.

📌 The benefits:

1. New employees get a better experience with everything set up for their first day.
2. HR and Tech emails aren’t forgotten or lost in the inbox.
3. Humans can focus on the important human elements of the onboarding process with basic pre-boarding tasks handled by AI.

This is humans + AI working together.

AI agents are (probably) a big part of the future of how we work. Mass adoption will take time though, especially in the world of HR and L&D.

Idea 3: Real-time Course/Skills Coach

In practising what I preach, I use real-time AI coaches in some of my courses.

My Art of Performance Consulting course, for L&D pros who want to stop being a McDonald’s drive-thru and create meaningful change, has one of my most popular coaches.

You can learn why I built this and how I did it in my full review on the blog.

The problem we’re solving

The cliff notes are I embedded a conversational assistant to help students put what they learned into practice.

I’ve constantly heard in my career how people struggle to put what they learn into action. I think technology can help us here.

It acts as a sort of assistant teacher (if we can call it that). Supporting students when they need it most to enhance their experience.

It also helps as:

• In the flow practice partner
• Review feedback from conversations and events
• Remind students of tools, frameworks and how to use them
• A private space to try new things

I find it’s used across all the course content.

Students are curious to unpack their specific context and examples in real time. I see a lot of potential for AI to support this, and that’s why this assistant exists.

The feedback from students has been great too, with high levels of daily active users even after course completion. My goal has never been to create ‘one and done’ experiences, and AI coaches help me keep students at the top of their game.

What are AI agents? (For Non-Techies)

I’m well aware, on idea 2, I dropped some words that might sound like an alien language to you.

So, let me backtrack a bit here.

Right now 99% of people use straightforward AI conversational tools like ChatGPT. That’s great. Some even build basic AI assistants in the form of GPTs – also good.

This is just the beginning.

Lots of big companies are working with teams from Microsoft, Google, OpenAI and more on their next-gen AI agent tech (yes another confusing mouthful).

Assistants vs Agents

Two terms you might hear techies mention with AI products are ‘AI assistants’ and ‘AI agents’.

Here’s the difference in clear, simple terms.

Let’s start with what we know – AI assistants like ChatGPT. These are tools that help us with tasks through conversation. They can write, analyse, explain, and give suggestions based on what we ask.

AI agents take this a step further.

Instead of just helping through conversation, agents can actually complete tasks on their own. They follow instructions, use different tools, and make basic decisions to get things done.

The key difference is simple:

• AI assistants help you with tasks
• AI agents complete tasks for you

Both are valuable, but they serve different purposes. An assistant works with you through conversation, while an agent works independently based on your instructions.

Use this info to impress the boss at your next meeting.

I’m not going to leave you with just this, though. As I’m a tech nerd, I’ve filmed a quick video to show how agents work with examples from Google and Salesforce – enjoy.

Useful Tools & Resources to build with AI

I’m keeping these super simple with assistant builders only:

• ChatGPT GPT Builder – you need a minimum paid $20 account to access this feature.
• Sana AI – friend of the newsletter, Sana, has a beautiful AI platform where you can create AI assistants on their free plan. It comes packed with free templates, and a whole host of useful features.
• Gemini Gems Builder – if you’re a paid Google Gemini user, you get Gems instead of GPTs. Same thing, different name.
• Chipp – the platform which runs a few of my assistants. You can use it to create free assistants or upgrade to a paid plan if you need data protection for work.

Plus, for ChatGPT users, I have a zero-cost “How to build a GPT assistant in 1 hour” course. Oh, and before you build any AI-powered product, get clear on why and what problem you’re solving.

---

Before you go… 👋

If you like my writing and think “Hey, I’d like to hear more of what this guy has to say” then you’re in luck.

You can join me every Tuesday morning for more tools, templates and insights for the modern L&D pro in my weekly newsletter.