10 Hidden Microsoft Copilot Features You Need to Know transcript

Millions of people now have access to Microsoft Copilot through their Office subscription. But a surprisingly large amount of people don't even realize it, and they don't even know how to use it properly. Maybe opened it once, tried it, got a flat answer, and never came back. Here's what I tell you now. Copilot has actually gotten really, really good in 2026, and a lot of people probably aren't using it to the best of its ability.

You can now ground it in your actual work. You can choose different models. It already lives inside a lot of the apps that you use. And the new co-work feature has really brought the agenda capabilities way further. But as I said, most people with the license haven't touched any of this. I've been using Co-Pilot in my work every day for the past three years, and I've seen just how good it's gotten. So, if you're new to Copilot or just want to get more by the end of this video, I promise you you'll go from barely touching it to knowing exactly how you can use it to the max of its capabilities.

For those of you new to the channel, my name is Ken and I'm the AI productivity coach. And I've taught dozen people how to use AI to make their work and life way easier. here. So, if that sounds like you, please make sure to subscribe and join me on this journey. So, let's run through Copilot's many features. Starting off with the beginner features. So, the first problem a lot of people come across is that they ask Copilot a question and they get a generic answer and then they effectively give up on Copilot.

The reason this might be happening is that you haven't grounded Copilot yet. And once you ground your work, it becomes five times more useful. Generally, with these AI tools, I think having the right context is the most important thing you can have right now. an AI that knows who your manager is, what you're working on, who's emailing you, all the context on the things that you're working on, all inside a safe, approved environment that gives you way more personalized answers than a generic chat.

And so when I say grounding, I basically mean work IQ. And so you can see work IQ is up here in the top leftand corner. Work IQ basically gives C-Pilot access to see your emails, your teams chats, your meeting transcripts and files. basically lets C-Pilot understand your RAW and your projects instead of treating every question like it's coming from a stranger. Just make sure this is switched on so you get better results.

This to me is one of the best features in Copilot and it's really improved my answers since I had it switched on. And so to quickly show that you can basically do a forward slash in your chat and then type the name of a file, a meeting or whatever it might be. And so in this case I am looking for my weekly priorities doc and I can tag it in right there and you can see copi can now pull from the word doc. You can see also that you can tag in meetings.

Similarly you can tag in different people from your orgs, emails, chats, channels. You can basically tag anything in the office 365 environment. So this is a way better way to pull in context than just describing your file or copy and pasting it. So the second feature to help you get really ahead of others is custom instructions in the personalization section. So, you can go down to the bottom left here, go to settings, click on personalization, and then you can edit custom instructions.

And basically, custom instructions are instructions you can set once that apply to all of your chats. So, for example, I like copilot to be very direct with me. So, I wanted to lead with the main point in the first sentence, be concise, clear. Use plain language unless technical precision is required. Now, you can configure these instructions however you want, but it's a great way to make sure the answers are aligned to how you like to receive communication.

Another personalization feature that is really helpful is saved memories. So you can turn on save memories and copi will basically remember details about you to provide better responses based on and you can manage the save memories and delete them as you like. So this is a really quick thing you can set up especially after you've just grounded your co-pilot experience. With this feature, it takes copilot from just answering questions to answering questions the way you actually want to be answered.

Another really cool feature of copilot is that it exists in your apps. So you can think of PowerPoint, Excel, Word, even Outlook. So this experience that I'm showing you now is m365.cloud.microsoft/hat. But you can access Copilot directly in these apps as I said. So you could simply do a prompt to get you started in one of these, like for example in PowerPoint, create me a presentation on the benefits of using AI to improve productivity.

So this is a really helpful feature if you work directly in these apps. Maybe you want to make a specific change on a slide, help with some formula. One area that I find it really useful is in Outlook. Sometimes I prompt it something along the lines of look through all the emails I've received in the past 90 days and flag any subscriptions or emails I should unsubscribe to. So, this makes Cop really good as it isn't separate tools that you need to remember.

They're sitting in your normal workflow. And this is why a lot of people miss it because the button is actually quite small, but definitely worth trying out, especially if you work in any of these apps already. Moving on to more intermediate levels of how you can improve your co-pilot experience. And the first one is that you can actually choose the model in your copilot experience. Most people just assume that copilot operates on one model.

The benefit is that copilot has the standard models that it uses but also has access to certain clawed models and has access to some of the open models and even GPD 5.6 was released a couple of days ago and co-pilot already has it in this experience. This is great because different models genuinely have different strengths. So depending on your workflow, it can be really helpful to switch between them and get the best out of each model.

The next feature that's really cool are agents. Grounding gets really good one-off answers, but if you need to repeat the same rules every week, this is where agents become really good in Copilot. So you can browse through different agents that are available in your org or that Microsoft has created or potentially other businesses have created to help you. Or you can create your own agents. It's really easy to create an agent to go into create agent and you can basically describe what you want.

So, for example, help me create an agent that responds to my emails in my tone of voice. I like to be very brief in my answers. I also like to lean on data. I want you to create an agent that helps you respond to the emails and also formats it in bullet points. Once you have your agent created, it will follow the rules and instructions that you've given it. And you can tag the agent into any chat. So, for example, with my follow-up email assistant, I could tag in the agent, say, could you respond to my email? tag in the email and then the agent should write the email based on the instructions that I've given it.

Agents are really good for tasks that are repeated on a frequent basis. So if you have a task that's repeated more than three times a month, I think it's worth building an agent for this task. The next really great feature for an intermediate level is notebooks. And this is kind of like Copilot's equivalent to Notebook LM. If you've used Google's notebook LM, this will feel familiar. You gather a curated set of sources and the tool only answers from those.

So basically, it's a great way to limit the amount of context you have and get answers or prompt copilot with those sources alone. Copilot notebooks is pretty much the same as notebookm except it can be grounded in your work files. So you can see here I can basically tag in copilot chats or add different references. You can add up to 300 reference files and then once you have all your files in there, you can basically prompt it like it's a normal co-pilot chat, albeit it's limited to your context.

Within notebooks as well, you can tag in meetings, chats, and all those things as well if you like. And if you like, you can actually share collaborators to your notebooks. Your chats will remain private unless you explicitly share them with people in the notebook. So, a great way to collaborate with someone on a project or a client, as I said. Finally, moving on to one of the more advanced features I mentioned at the start of the video, and this is co-work.

And this is where copilot stops just being a chatbot and moves more into that agentic experience. Some tasks are a whole sequence spread across a week, pulling from multiple sources, sometimes producing more than text. Normally, that's a recurring chunk of your calendar, which is done manually. But this is where co-work changes the game. It plans and runs multi-step tasks across your M365 apps in the background and returns a finished deliverable, not just a draft.

One thing to note is that your IT admin might have to enable it as it's built separately based on usage. One of the core features of co-work is that you can use skills which are basically instruction files. There's some built-in skills that Microsoft already have like PDF, Word, Excel, and that basically allows co-work to create and edit a lot of these files. So, it makes it really good for working in these Office 365 apps.

But you can also create skills yourself. So, basically, you can chat to co-work about what you want and create the skill that way. So maybe it's again a task that you want done. Could be, for example, you wanted to check your emails, create a to-do list for the week and then get co-work to start actioning on some of these. So maybe you'll have to prepare a report and so you can get co-work to do this based on the skill that you create.

One example could be you want a recurring weekly scan. So you can see here you can also create scheduled tasks. So for example, a weekly priorities list that runs every Monday as an example. There also plugins you can install as well. Also, for example, fabric IQ or dynamics if you work in that space. You can also add approval steps in co-work as you want. So, for example, rather than it sending something on your behalf, it stops and asks for approval first.

So, co-work really moves co-pilot from a chatbot to an actual agendic experience. Most people never even touch the scheduled task or the skills section of co-work. Once you get the hang of that, that's when you truly get the value out of Copilot. So, that was a really quick run through of all the features that I've seen make Copilot go from a simple chatbot to a really great AI assistant. If you start using these features, you can quickly move from a beginner to expert and really increase the quality of your work and save yourself a lot of time.

I definitely think spending time up front with custom instructions, turning on your memory, turn on work IQ, getting your agent set up really adds value down the line. Then, as you move down the line, you can start incorporating more things like notebooks and co-work to really scale your impact. Please let me know in the comments, have you started using Copilot or what do you think of it? If you found this video helpful, please make sure to like and subscribe and join me on this journey.

I also have a weekly AI productivity newsletter and you can subscribe to it using the link in the description below. Thank you so much for watching. I really appreciate it and I'll see you in the next