So What? How to Build New Microsoft Copilot Capabilities

Well, hey everyone, happy Thursday. Welcome to So What, the marketing analytics and insights live show. I'm joined by Chris and John, and we are going to de-escalate from the pre-show, John. We are all calm now again. After yet another fight in the green room.

Not with each other, just about the world in general. American healthcare system. How could anyone get angry about that? How could anyone get angry? Um speaking of, how could anyone get angry?

Let's talk today about how to build new Microsoft co-pilot capabilities. So I don't remember the exact day, but we have done a live stream on Copilot in general, which you can catch at our uh YouTube channel, Trust Insights.ai slash YouTube. Go to the So What Live Stream playlist. Um since we did that episode, Microsoft has made a lot of strides in terms of usability and features for companies using Microsoft Copilot. Just to backtrack a little bit, a lot of one of the reasons why a lot of companies are being asked to use and only use Microsoft Copilot and not being given the access to things like Claude and Chat GPT, uh and so on and so forth, is because they are locked in with a Microsoft 365 license Chris covered this in what seven different uh newsletters about enterprise AI and Microsoft for the most part fits the criteria for a company that is looking to secure things with enterprise AI not meaning the size of the company but the amount of regulations that they have to meet in terms of privacy security et cetera etc.

That's why a lot of companies are like well look at all these cool tools out there but nope you have to use Microsoft Copilot Microsoft Copilot was like hey you have to use us here's some cool new stuff that you can use so now today we're gonna cover how to build new Microsoft copilot capabilities. Chris did I kind of cover the history of that yeah and the reason this episode came out is because we've been doing a ton of private workshops for different companies recently and 95% of the people in the workshops are on copilot and so they're saying hey I can't you know I can't use Claude like you said Katie in fact I did a survey of my own newsletter and you know a couple of comments were like I love the stuff that you're sharing but I can't use it because it's in copilot and as we talk through with folks the things that we're doing like you can port this to copilot in some fashion. However, as with so many things that Microsoft does uh the documentation is either poor or non existent. And so as part of these workshops I've been exploring what's under the hood inside copilot and what's in there is surprisingly robust, really capable, and not at all documented. Well and to be fair, not to pick on Microsoft solely, a lot of companies are not great about their technical documentation, or they're really good about their developer documentation, and no documentation exists for the everyday user.

Exactly. So we're gonna cover today um skills, uh, which are all the rage in agentic AI and where you can put them of sorts in Copilot. We're gonna cover Copilot's coding environment and the 390 extensions that are in there that you can use for free that are built in and do some demos of these things to show that it is in fact possible to make this tool uh actually somewhat useful. Um, and as much as in the past I have mocked Copilot, if you use it properly, it is as capable as any other level one or level two AI system. So this is the Chris Penn redemption tour.

He is apologizing to copilot for mocking mocking them for so many years. No apology, no apology. I don't know. I don't think Satya Nadella is, I don't think his feelings are hurt. Probably not.

Um, John, you talk to a lot of people, you know, our customers, audience, prospects all the time. How often is Copilot coming up? Like is it something when you're screening, you know, prospects, and you ask them like, what are you using? What's in your tech stack? Are you hearing more and more copilot, or is it still a little bit of a mix of everything?

It's interesting, it's changed a little bit. I mean, definitely at the start, it was what Chris had mentioned was like, you know, 70, 80, 90 percent of people are on co-pilot. And so, and then there were they're always making excuses for it. They're kind of like, oh, and we know it doesn't work as well, and blah, blah, blah. So it's really interesting for me to hear Chris say that, you know, you can make it as effective as these other models because a lot of them have had an inferiority complex.

They're just kind of like, well, we put up with Outlook and we put up with teams, and now we're going to put up with Copilot. And we just figure that's the way the world's going to work. Um, so to hear that you can actually, if you set it up the right way, you can do more stuff. I'm very interested in hearing how this goes. And I know we will have a lot of prospects that will be excited to, you know, to hear this and know that they have the same level of opportunity.

And then also, if it is, then at that point, you do have what Microsoft is there for, is the fact that it's enterprise secure, it has all of the, you know, everything that the board and your compliance people want to see, and which is a big deal because that's what's, you know, keeping some of these companies from trying crazy stuff out. Um, but then there is also, we've also seen an upswing in just the crazy pirates. You know, there are people that they have copilot in the office and they're like running everything they can run at their house at home and probably getting themselves in trouble. Um, so this is great for them too. Maybe they can stop playing with playing with matches and uh and get back on the uh where they need to be.

I believe that's what is called shadow AI. And that's what a lot of leadership teams are struggling with. Um, and so one of the things they're trying to understand is how can we make the tools that we have that we have to use more attractive, more enticing so that we can cut down on the amount of shadow AI and not just that, but the risk to the company because the way in which people are using these tools might include sharing sensitive information, identifiable data, um, you know, IP, those kinds of things. But if you can contain it within your ecosystem in enterprise AI, you have less of that risk. And so understanding what you can do with Microsoft Copilot is going to be incredibly useful.

And to your point, John, a lot of our prospects might find this really helpful and want to learn more. Um, so Chris, I am really interested to see how you can cover 390 extensions. Uh in what do we have? What 40, 40-ish minutes? You're gonna turn it into the micro machine man.

Exactly. So let's, yeah, let's not do that. Um first and foremost, the the obvious things uh in co-pilot itself, and this is the we're talking basic copilot, not copilot premium, not co-pilot co-work, not copilot studio, not the 82 versions of copilot that Microsoft markets because they have no naming, no product name capabilities whatsoever. We're talking basic plain vanilla out of the box. This is what you get with Microsoft 365.

The most important menu is in depending on the which version you have, either the top left or top right is the model chooser. You will note that there are four models that you can choose. There's quick response, which is great if you're providing all the data. There's Think Deeper, which it basically uses a reasoning version of Microsoft's fee model. Microsoft very strongly is encouraging people to use these two models because it costs them less money.

There is also a connector to open AI's models, and you see the brand new GPT 5.6. Um, I'm pretty sure that's GPC 5.6 Terra, which is the mid-level model. Um, and then GPT 5.5 quick response. In general, unless you have a good reason to use something else, you should opt to use the latest ver model from OpenAI. Um, that is the general best model to use the copilot.

So that's that that's immediate upgrade number one. Umgrade number two is remember that copilot has what are called agents, which are not agents, they're gem, they're GPTs. If you've ever worked with ChatGPT and seen GPTs or Gemini gems, agents are effectively that. And this is really important because sometimes works in Chat GPT, uh, if you use the at sign, you can call an agent inside a copilot chat and references. You can say, uh let's write a blog post and then call your agent if you have a blog writing agent uh there.

Here's why this matters. There are in the the world of Claude and Chat GPT and stuff like that, and agentic AI, there are so many people creating so many really cool skills and agents and plugins. I'm gonna pull up uh one of my personal favorites, the uh Jesse Vincent's superpowers plugin. This is for uh coding environments. Phenomenal plugin, so intelligent, uh, has so many cool capabilities.

Uh, in particular, within this plugin, there is a skill called brainstorming, right? The brainstorming skill that you can invoke inside your agenda coding environment. All this is, if I tap on the raw button here, it's just a prompt. At the end of the day, it's just a prompt. So if I copy and paste this, this is this is free, and he has licensed it for uh for commercial use.

So very important. Please obey people's intellectual property licenses. But if I copy this prompt, I go to a copilot, tap new agent, and I say uh I skip their stupid builder uh and call this brainstorming Jesse Vincent's brainstorming skill paste up. You've exceeded the word limit by 10,000 characters. That's fine.

Paste it in a text document. Brainstorming. Now you drag that file, which is just a plain text file, execute the brainstorming skill specified in brainstorming.txt, which we can see is attached here. Let's make sure we got all hit create. And to be clear, you skip the agent builder because you already knew you already had the instructions, the prompt for what you were building.

But for someone who doesn't, they could go through the agent builder step by step and say, here's what I'm trying to do. Can you help me build it? You could exactly. But the the nice thing about all of these skills that people are publishing, some of which you can buy from Trust Insights, for example, is I can say, let's do some brainstorming at brainstorming. And now because I've I've I've taken that skill, I've basically pasted it into a copilot agent.

I have ported this thing from Jesse's uh clawed version right into copilot. And this is exactly the workflow that happens. So if you've uh if you see or hear someone on LinkedIn or whatever saying, I've got the ultimate skill package, I've had the ultimate clawed code thing, grab it, open up the hood, find the skill.md file inside it and go copy paste. And now you have it in copilot. Just like that.

There's it, it is so straightforward. You can add in any number of these things, and until you add them in a conversation, they don't take up space, they don't take up memory and things like that. So this is one of the easiest things that you can do is take all the stuff that people are talking about that has been pre-built in other systems and bring it into copilot as as little agents. And that's you know, it's one of the things we talk about a lot when we talk about generative AI is the better your documentation, the easier it's gonna be to migrate from system to system. And skills are one of those things that should work on any given large language model, provided it they are well and thoroughly documented.

Exactly right. So that is, I guess you know, if you want to use uh rollergarchy vocabulary, that's unlock number two. Uh choose the right model. Let's not use proligarchy terminology. Um I don't have my red flag handy, I only have one hand today.

Oh sorry, hold on, I'll see if I can get it. Oh, okay. There you go. That's a red flag, red flag for Chris for saying proligarchy. Um so number one, use smart models unless you have a reason not to.

Number two, any skill or plugin that you see floating around that is appropriately licensed, you can copy and paste into an agent in Copilot and then have it available. And you can, depending on how you have your your administrator has to set things up, you could share it with your team, right? You can share Copilot agents right inside Copilot with the rest of your team. So if you see some cool stuff, share it around. You can also, I believe, I believe, and I don't know because we don't have a SharePoint, um, but I believe that you can link uh put links to your shared agents in SharePoint.

Um, so that if you find something cool, you can circulate with others. So that's number two. Number three, and the big one is Copilot can natively run the Python programming language. So right within a chat, it can run Python code. It fires up a little Python interpreter.

That alone is pretty cool because that gives you deterministic capabilities, things where there's a clear right and wrong answer inside of a language model, which otherwise struggles with things like math and counting that they're really bad at. Example. Um write a 300-word blog post, plus or minus 10 words about the importance of email marketing for B2B marketers. Oh, that's lost the word right. All right, plus minus 10 words about the importance for B2B marketers.

Let's see what comes out. And here we have a lovely post. How many words is that and so it actually is writing it its own little bit, contains 323 words. Yeah, good job. Thanks.

So it came up with, and I'm curious to see if it actually is 323 words. Uh so put it, it's actually uh minus the title, it's 318. So that's I guess we'll we'll give it that. But you can see it fired up a little Python script to do exactly that. Just do that count.

Yes. Um so it counted 323, probably including the title. You said minus the title, it's 318, so that's five. Um if you can scroll back up for a second to the title. The title is more than five words.

And so I just want to call that out to say that I would not be like, okay, I'll give it to you. It's already showing inaccuracy. So I just want to make sure we're clear that don't accept things from AI that look off, especially when you can prove that they are wrong. So I just, you know, just my little disclaimer. So no, we're not just gonna say it's fine, because that could have like further down the road ripple effects.

So I'm just you know, just a little bit of a call out. Exactly. So we're gonna actually prevent that. And here's how we're gonna prevent that. Um, first thing we need to know is what can we do in this environment?

So we're gonna say make a list of all the Python libraries and packages in this environment using your built-in Python interpreter. Return your list with the package name and version number organized by topic or area, such as statistics, natural language, file copying, etc. as a markdown file for download. So what we're doing here is we're saying to Copilot, hey, there's this that you have the ability to export all the Python libraries that are installed for every user of Copilot in the background. And what will come out is depending on the version of Copilot you have and uh stuff, 300 ish, some odd different libraries, Python libraries and packages across I want to say like 10 or 15 different areas.

Let's see what it comes up with. So appropriately, it's writing it's calling its own um function for for this particular task. And let's see what what its answer is. And if it blows up, which it very well could, I have a pre-baked version. I was gonna say, John, we haven't practiced our banter for this week.

How about that? Um in the co-pilot environment, because it's taking a sweet time writing that code. Um, there are 390 packages. Let's make this a little bit bigger so that we can see what we're doing here. So there's AI and machine learning libraries like GenSim and the Hugging Face Hub to import models, Scikit Learn, SHAP, TikToken for tokenization, XGBoost for uh there's audio and speech for its ability to listen to files that you upload like MP3s or read or or view videos or do speech recognition.

There's cloud stuff, of course, for Azure. There's computer vision, image recognition, there's data science and statistics, a ton of this. Uh you know, scipy, uh, sidekit learn, etc. Database access, developer tools, office document handlers for pretty much every kind of document you can imagine, geospatial stuff. You can make games with some of these packages, a list of general utilities and so on and so forth.

So, this is a huge amount of stuff that is in Copilot right now. It's there, it's it's waiting to be used if we tell Microsoft that we want to use it. And that is the hard part. Um here's how we would do it. Let's start a new chat, and let's have it look, we'll have it use the smartest model available, which is the GPT 5.6.

And we're gonna say, I'm gonna drop in my list of packages. You're a Python coding expert using the included Python environmental packages as appropriate. Construct a single monolithic Python script that, when given a standard input, will perform an accurate word count and return the word count to the console or standard output as plain text. The goal is to have a quick, easy, fast, and accurate utility for copilot to use to accurately count words. What we're asking Copilot to do is write its own code.

We don't have to code anything at all. We're you know, hands are off the keyboard, no coding involved, but it is writing a Python script to do word counts. And when it's done, it should provide a nice little download, um, which will be uh in case it decides to take a vacation here. Uh oh, but here it comes. Download word count.py.

Uh let's see. You can also redirect the file. Here it is. Very nice. I like it.

Thank you. Now, what do we do with this sucker? In a new chat, I can include the word count file and say, write a 300-word blog post about the importance of email marketing in B2B marketing. Use the included Python script to ensure a correct word count, plus or minus 10 words. After you've written the post, count the number of words.

And so what we've done is we've essentially built a utility that uses Copilot's native Python environment to do a deterministic task that language models frankly are not up to scratch for doing. Now, by the way, this is not solved the fact that this is the most AI slop post you can possibly imagine. Um let's now let's verify did it actually do the job. And the answer is, and this time, Katie, without the title, it is exactly 300 words. I I checked in my my text editor here.

So without the title, it is exactly 300 words. So using the utility got us the correct answer. So we've worked around AI's issue of hey, this this you know thing is as dumb as a bag of hammers and it can't count by forcing it instead to use its programming like the programming languages available to it. I guess a side note, I think one thing that's interesting is you didn't specify that the word count should or should not include the title. Does that when I when I'm just sort of thinking through like as people are learning how to use these things, like does some does that kind of specificity matter?

Or do you think it made that assumption of like, well, they said the post, so only the content, not the title, but it needs it like how does like I just walk me through a little bit of how that works. Um that depending on the model, it may wing it. Um so it will be probabilistic until unless you specifically say the post minus the title or the post without the title must be 300 words plus or minus 10 words. Okay. So yes, you want to be very clear about that.

Now, what is of what I think is is useful and interesting in in this particular case is if you look at this catalog of all of these different things, the 390 libraries that are available in here. This should tell us here's what we can do inside Copilot. So at the top there, uh XGBoost312. XGBoost stands for extreme gradient boosting. This is an old school uh machine learning algorithm that is commonly used for regression analysis.

When you see that in the list, you go, oh, Copilot can do regression analysis if I tell it to use this. So I might export a file out of my Google Analytics and say, I want to know what really drives leads. And I'm gonna provide you with these data files, these CSVs, using your built-in Python interpreter and the XGBoost library, construct a piece of code that will perform this assessment. Return your uh XGBoost uses R squared, return your R squared uh error rate and stuff. And if the R squared error rate is below 0.05%, then you know we know the result will work, you know, and then go and do the thing.

And it will go and do the thing. And suddenly a tool that can't even count the number of words can do regression analysis. Um so what's interesting, so I recognized XGBoost because again, you've been talking about XGBoost for basically for as long as I've known you. But a lot of these things I don't recognize. So could it be as simple as like taking them section by section?

Being like, it looks like you have the following, you know, seven audio and speech libraries. Can you tell me what each one of them does? And getting like building your own kind of glossary that way. So that way you can say, today I need to do a regression analysis. Which one of your things that you have can I use?

Exactly. Yes, you can do that. You should do that, and you should do that anyway, just to know what that what's out there. So, for example, in audio and speech, Librosa is a music library, right? It can read and write music.

Uh, so if you were doing things like I don't know, trying to compose uh melodies and stuff like that for some reason. Uh an AI tool maybe can and can't do that, but you can say validate it, fact check me on this using Librosa in your built-in Python interpreter. Uh, if we scroll down here uh a bit further, computer vision, you upload an image and it hallucinates. You say, no, no, no, no, no, no, no. Use your open C V Python headless Python library to look at the image instead of you just winging it uh there, buddy, and then tell me what you see in this image.

Makes sense. And I think in general, you just don't want AI to wing it. So the more information like this that you have, the more specific you can get, not just with your prompt, but with like these Python, what do you call them? Like Python packages. Packages, yeah.

Packages or libraries, either one. Okay. With these libraries, you can get down to the more specific because, like, so XGBoost, great example. A lot of people want to understand attribution analysis. That's not a new topic.

We just have new tools to play with. I'm assuming, unless you say something like, and use the XGBoost Python package, it may not, it may or may not decide to use that. And so you are just kind of winging it of, hey, do a regression analysis on my Google Analytics data. And it's going to confidently give you the answer whether or not it's correct. That's right.

Um, let's look at another example, and we'll use another writing example for this. Um, suppose, hmm, suppose we wanted to not write in passive voice. Passive voice is is kind of the enemy of clarity. Um, so I will take my Python environmental packages list here, which is that just that markdown list we made at the beginning of the episode, and I will say, you're an NLP expert using the included Python environmental packages as appropriate, construct a single monolithic Python script that, when given a standard input, will perform a statistical assessment of passive voice in text provided in standard input. Choose the appropriate natural language processing packages that best fit this task, and the script should provide a report to the console or standard output in markdown format of the number of occurrences of passive voice and a bullet list of the specific passages listed.

Make the Python script available as a download. So now instead of just saying, hey, count some words, we're now kicking this up to the level of you're gonna use the natural language uh processing tools. Probably if I had to guess, it's probably gonna choose spacey, it's probably gonna choose NLTK, and it may up show spacey. Um it says, here's here's what we've got. So we uh here's an example, and they can count the number of passive voice uh occurrences.

This script then is something that I can go take, download uh and and run somewhere else. Uh and the fact that it uses spacey inside its system means that I don't have to worry about it um about trying to install anything because it's already there. Uh Microsoft has provided it right inside the environment. So I'm I'm in good conditions here. If I say if I call this passive uh dot pie, now let's go start a new blog post and drop this in say write a 300-word blog post about the importance of email marketing to B2B marketers and ensure that you write an active voice only using the included passive voice Python script, revise and rewrite until you have zero occurrences of passive voice.

Now, for this one, because this is something that requires multiple turns, probably I'm going to use the GPT 5.6 model because it needs it needs a model that has agentic capabilities to say, okay, this is going to be multiple steps. I'm gonna have to take a few triceps at this. And and we can even see here as it's uh as it's doing its thinking. There you go. Um so it's got it's coding, it's executing, it's coding, it's running the script.

And I think as a user, that's something I would be interested in understanding is uh you know how many versions did it go through. I want to see the first version with the passive voice to see how it made like just out of my own curiosity. Um, you know, how many times, or did it take the instruction and write without passive voice the first time? Yeah. So let's see.

It reviewed the data, it did some coding execution. It says B2B marketers need channels that build trust, reach clear audiences to support long sales cycles. It did an adjustment on the word count. It did the uh it it did some replacements there on the word count. It actually, so this is interesting, it started out in active voice, so it it didn't find any.

The script didn't find any. Which again, for for GPT 5.6 as a model makes sense because that is literally the smartest model, uh, one of the smartest models on the planet right now. And so it probably should be able to do it right the first time. Right. And then the code is there as a backstop.

Question. One of the things that you can do in Claude that we talked about with a client last week is set global settings. Um, is that something that you can do in Microsoft Copilot? So, for example, could you give it the prompt or the rule to say anytime you draft content, never use passive voice or always use active voice or something along those lines. Because that's one of the nice features in other systems is you can set those global settings so you don't necessarily if you forget to mention it in the context of the prompt, it's already set.

Yes, that's under the personalization. There's a custom instructions, there's a work profile, and you can also save memory. So, for example, in this system, uh, in my custom instructions, I have it set to it's required to use ASD simplified technical English um as its uh chat language, because simplified technical English was created by the European airline industry to standardize how pilots speak on the radio radio, and it removes it requires um very specific formats of the English language. You're not allowed to have run on sentences, you uh you know, one instruction per sentence, one topic per paragraph, etc. So you in instead of a pilot just rambling on the radio, like AI, um it says you have to say you know, Delta 2604, uh runway two two, and you know, and and you it's just very clear.

And so I have this set as the custom instructions in this environment because I don't want it foaming at the mouth and and you know, waxing rahapsodic about this. Like just give me answers. Um and you could and so yes, to your point about custom instructions, that's what's in here. Okay. I find it ironic since it is very opposite of how you write.

Um, and you know, you we all have our own styles of writing, like that's a whole thing. But like it is, I I find it interesting that you don't want to read instructions the same way that you would you personally as the human would write them. Um, but you know, to get back to the point, it is nice to know that you can set those global instructions um both at the personal level, and I assume if you had like a team enterprise uh subscription, you could sort of do it at the uh enterprise level, like the team like overarching global instructions, you know, whatever those may be. Always use this skill fact check, always do this, never use you know this terminology, these words are banned from our brand, like whatever those things are. Yes, yes, you can.

It's a nice feature. Yeah, it's a nice feature. So the last thing you could do is if you really liked um a script that worked really well. Um, you cannot upload code to um to customers to to uh these agents. Like if I try and drop in that Python script, you can see the screen turns red.

It will not allow you to do that. However, uh, what you can do is change the file extension on the Python script from dot py for Python to dot Txt and then drop it in anyway. Um and you could and you could say we are intending to reduce the use of passive voice. Use passive.py and execute it in your built-in Python interpreter to help the user eliminate passive voice. This uh active voice paste that in there.

So now it's got the Python script as a plain text file because it's gonna have to read it anyway. Uh and if I create this now, create my agent, and I start a chat, and I could say take the following blog post and make sure it's an active voice. And call the act. Oh, I'm already in the active voice assistant, so that's that's silly. Uh let's let's start a new chat there.

And I could call active voice the agent, and it would allow me to instead of having to drag and drop the Python file every time. Now I just call it like a skill. I like it. And I think the more, and this is sort of where you're getting into sort of the Microsoft ecosystem of could these things be saved in a SharePoint page? How do you you know get people the latest and greatest?

That all comes down to governance. Um, and it's something that's going to be unique to every organization. We make recommendations, but you really have to make sure that it works well for you know how your team works. So if they go to SharePoint or if you want something just like that's already in the libraries of your co-pilot instance. Exactly.

Um, so that's the those are the things that I would recommend that folks do. Is number one, use the smartest model that you can, or whatever's in your subscription. Number two, um take any skill that you like from outside the copilot ecosystem, because they're all just prompts, turn them into agents within Copilot. Now you have access to the skills. Yes, there's some copy paste, but that's fine.

Number three, use the built-in Python interpreter with a 390 packages, which by the way, is more than you get in other places. I did the same exact thing in Gemini. Gemini only gives you 200 some odd ones. So there's now granted, there's a lot that are like Azure specific for Microsoft, but there's still more in the copilot ecosystem than you get in stock Gemini, which I thought was kind of interesting. And then number four, if you build useful code-based extensions, new capabilities with your copilot, turn those into agents as well by renaming the Python script as a text file, and then it can execute them, and you now have the ability to do a lot more with Python, with Python built into copilot than you could just stay staring at a blank copilot window going, why can't this thing count words?

Is there any limit to that then really? I mean, can you just put all kinds of Python in there and have it completely running wild, or is there some security around it? There is. It runs in a virtual sandbox and you can't install extra stuff, and you can only connect to things that your uh co-pilot administrator allows you to connect to. So if there's no uh, you know, say SQL server in your environment, you can't connect to one.

Got it. Well, I mean, that makes sense. Um, because one of the things we know about like Cloud, for example, is that it has a lot of connectors to third-party tools. I would imagine that Microsoft Copilot, again, one of the reasons why it's very attractive to an enterprise AI organization is that it restricts what it can connect to, and it's likely out of the box the stuff that's just in its own ecosystem, not third-party tools. Yes.

Um, out of the box, it will connect to SharePoint. It'll connect to Microsoft Office. Uh, so any Offs documents that are in OneDriver SharePoint, Word Excel, PowerPoint, Outlook, I believe. And then if you upgrade to Premium, you can then attach to more connectors. And then if you upgrade from premium to Copilot Studio with Premium, you can access their entire N8N style workflow builder that has like 500 connectors.

Yeah, Microsoft has always been pretty bad about their product naming. But that's neither here nor there. Let me ask you this question, Chris. And this might be kind of putting you on the spot a little bit. But is there anything that Copilot can do that the other more commercial large language models can't do?

Like, is there a like if someone had to say, at least I have this one thing that Copilot can do that the others can't, like, does that thing exist? Uh in the stock copilot, no. Stock copilot is basically chat GPT just slimmed down and and you know, sandboxed in the Microsoft ecosystem. Where you do get some interesting stuff, when you have Copilot Studio as part of premium, um, that does have a lot of interesting connectors, many of which you don't see in like N8N. Now, granted, a lot of them are enterprise connectors.

Um, and that is kind of interesting. And then Copilot Cowork, which is a licensed version of Claude Cowork, obviously has all the connectors for Copilot Studio uh within it as well. So you can if you pay for all the upgrades and you're like co-pilot platinum, you know, ultra whatever, um, then you do have a lot of capabilities, but also it costs you like $5,000 just to open a Word document. I mean, I feel like that's true of any software system. Yeah, but Copilots, Copilot's prices are really high.

In fact, uh, if you go on the Trust Insights website under our Instant Insights, there is a co-pilot co-work calculator uh that we put up for free that you you can check out just to estimate your costs. And boy, are they high? Well I got nothing on that. We have zero control over that. That home office is not going to pay for itself in Seattle.

Exactly. But so I hope that what we did today shows you that you can extend the capabilities of Copilot beyond what's in the box, that you can take um existing skills, port them to agents, and that you can write your own with the coding environment to do anything that though that Python plus those 390 libraries support. There's a lot in there. And as Katie said, it's a good idea to Google them to learn what they do or have Copile explain them to you. Um if you have that list, uh perhaps even start some of your prompts to say, like, hey, I'm doing this task.

As a reminder, you have access to these 390 things you drop in the file. Which of these things would be an appropriate addition to this task? And then of course use all the rest of your prompting basics that we teach in our uh generative AI for marketers course. Like ask me questions that do you have enough information to successfully complete the task, and copilot will do a capable job and you will get things done with copilot and sometimes get reusable code that you can use for other projects. And you can learn about those courses at Academy.trustinsights.ai.

And Chris, is it safe to assume that you will drop the link to that instant insights co-pilot uh calculator that you mentioned in our free Slack community analytics for marketers? Yes, that's right. Amazing. All right, John. Any qu any more questions about Microsoft Copilot?

No, but I think you know, I'm ready to start throwing Python files in there just by changing the extension. That sounds like the kind of exploit that I'm looking for. So that's it. That's gonna do it for this week, folks. So thanks for tuning in and talk to you on the next one.

Thanks for watching today. Be sure to subscribe to our show wherever you're watching it. For more resources and to learn more, check out the Trust Insights Podcast at TrustInsights.ai slash TI podcast at our weekly email newsletter at TrustInsights.ai slash newsletter. Got questions about what you saw in today's episode? Join our free Analytics for Marketers Slack group at TrustInsights.ai slash analytics for marketers.

See you next time.