HITT – How To Turn Microsoft’s AI Platform Into Real Business Wins for Your Clients

So moving into our presenter segment, I’m excited to introduce our presenters today because they’re going to show you exactly how to turn Microsoft’s AI platform into real business wins for your clients. Joining us, we have Chad, Chris Fish, Ford Luken from Mobile Mentor, as well as Denis O’Shea, our founder and CEO. Guys, thank you so much for being here. Chad, I’m gonna hand it over to you to kinda kick us off.

Great. Thanks, Cassandra.

Again, thank you from the Mobile Mentor team for joining today. I think it’s gonna be a fun conversation that we’re gonna have. We like these fireside chat type models to go through, and I think the topic of conversation today is appropriate.

And I really appreciate you the three of you joining me today to to talk this through. One of the key things that I hear a lot of times from our partners and customers is Copilot sucks.

And that’s one of the key things that kicks off a conversation in and around Microsoft AI. So with all of that being said, why, why do do they feel that way, And and how do we overcome that more more importantly? How do we overcome that mindset with the customers? And, Denis, I’ll I’ll kick it over to you to to start things here.

Sure.

Look. That’s a great way to start. And and the analogy I would give you is that if your customer is a hospital, the computer experience for anyone working inside the hospital probably sucks as well. And the reason for that is that people the the end users, the nurses, and the administrators probably compare that computer experience to what they have at home.

At home, they can just buy a machine from, say, Best Buy, set it up, use it. They have a nice easy user experience. In the hospital, that computer has been has been provisioned to be a totally different experience because you’re protecting PHI and PII in a very sensitive environment. So there’s probably twenty years of design gone into securing the data, securing the device, securing the access to the EMR and all the things, and it’s a completely different experience in an enterprise setting.

So just hold that example in your mind for a second. Now we go to AI. We’re all using AI in our personal lives. You know, you can just use ChatGPT in the browser.

You have a really neat experience. If you then try and transpose that to an enterprise environment, like a health care setting where there is PHI and PII, of course, you can’t just use naked, unsecured ChatGPT in an enterprise environment. You have to apply controls and security and some adult supervision to it in the same way you had to do with that computer to bring it into a clinical environment using it in a hospital. So the reason Copilot sucks and the reason so many of our meetings start with that language is that what Microsoft have done as the opening play in in the in the AI space is they took ChatGPT.

They added some guardrails, some security. They grounded the data in the Microsoft Office three six five environment. They provided some contextual relevance, and that changes the experience because they had to. Because if they don’t, there’s gonna be massive privacy breaches.

There’s gonna be huge security leaks. You imagine if Czech GPT was just running wild in an enterprise environment with no guardrails, you know, that hospital is gonna be out of business in minutes because the data is gonna leak. So that’s the reason why people perceive Copilot sucks. It’s because security controls guardrails has been applied to make it useful in an enterprise setting.

Does that make sense?

It makes complete sense. And and I’m seeing in the chat here and and this is great because all of you that are are watching this, we want you to interact in the chat and ask questions. And and Rick made that, same point here is is guardrails and data mining are a key factor in that. Yeah.

And and, you know, I’ve got other people chiming in that they don’t feel they can trust Microsoft with their data either. And I think I think that’s that’s something to overcome in general with AI, but specifically in the Microsoft and Microsoft and and ChatGPT world where those two have come together for Copilot. Yeah. And we can actually we can actually touch into, hopefully, down the road a little bit here in the conversation, how Anthropic is getting pulled into this now as well with with Copilot too.

Yes. But continuing down the path of of Copilot and implementation, can you can you kind of, kick things off with maybe a a a real world example of where you stepped in? You’re using health care as as an example. Maybe maybe give a a win or something like that that you’re working in that space right now where a hospital or a health system has been able to leverage Copilot correctly and and how you as Mobile Mentor go in and make that happen.

Sure.

We’ve got a a rollout happening today actually here in Nashville, one of the major health care providers. I think it’s about six thousand seats of of Copilot being rolled out to all the the nonclinical users initially, the revenue cycle people. And and the way it’s being rolled out is that it’s sitting on top of Microsoft three six five. That’s the first thing to understand. So all the data that people are accessing is enterprise data in Microsoft three six five. That’s in Teams, Outlook, OneDrive, and all the different Microsoft applications.

And and the first thing to understand about the copilot strategy is Microsoft strategy is not to try and compete and win the model war. Okay? Models will come and go, chat GPT, Claude, DeepSeq, all the rest of them. They’ll be the the models will come and go, and there’s trillions flowing into into that industry. And so we’ll see we’ll see a lot of movement in that space. Microsoft strategy is simply to be the UI for AI.

The UI for AI. They want to provide you a safe, secure user interface to use the model of choice.

So the rollout we’re doing at that unnamed health care provider, we’re showing them how you can select Claude, if that’s your preferred model, or select one of the GPT models if if you prefer that. So you choose the model you want to do the piece of work you want. And if I can share my screen, I’m actually gonna show you what this looks like because it’s a game changer when when you see the ability to select a different model.

Do I have And as as you’re bringing that up, I’m gonna answer.

There was a couple of questions that popped up here. So Mark asked, how do we sell AI to clients for real m m r or monthly recurring revenue opportunity? Just talking about it doesn’t pay the bills. Agreed, Mark.

And and the whole premise behind this is to leverage organizations like Mobile Mentor, and and others in the portfolio for Telarus to be able to get that monthly recurring revenue around offerings and support and ongoing development and operations DevOps on different platforms just like Denis is talking about here. The ongoing support, the training, all those aspects of things, it’s it’s the services built around these models currently, and that’s what’s happening here. So what is is a key factor is taking the conversation and turning it into that MMR is what what we’re driving for today.

Absolutely.

Yeah.

And I’ll come back to that because Absolutely.

We have a model now that is generating MRR, which we’re taking to market through this community. So we we get to that, but there’s a little bit I wanna cover before.

Absolutely. Please.

Can you see my screen, Chad? I can. Yep. You can. So for just seeing a basic chat in ChatGPT, you can choose Claude or chat oh, sorry.

In Copilot, we can choose, Opus Claude or one of the ChatGPT models. If we got a researcher, which is a phenomenal tool to go deep into something, they the the response is gonna take much longer because it’s doing deep inference. It’s doing deep referencing. But, again, we can go here and we can choose.

I want auto, just choose the best model for the task. Critique, which is fascinating. I saw this yesterday. So it uses GPT by default and then gets clawed to pressure test us and challenge the answers that come back from from GPT. Model council, I saw this yesterday. It creates two columns, and it shows you the GPT response in once in one and the Claude response in the other. So you look at both, and you can figure out which one is more appropriate for the work you want to do.

And then the The the co work model where you get, you know, multistep workflows, I believe, again, here, yes. We can choose some of the different cloud models because they came up with some of that workflow capability.

So the first key point is understanding Microsoft strategy is not to win the model war, work with the best models in the industry, and allow you to choose the one that’s best served to the kind of work you’re doing, whether it’s an Excel or PowerPoint or Word or whatever you’re doing. Choose the best model, and then be able to use it in a way that’s it’s the the work is grounded in your data, your teams, your OneDrive, your Outlook, and that it’s all complying with your security policies. So if you’ve if you have defined your data correctly, you’ve classified your data, you’ve maybe applied some sensitivity labels, you’ve got some data life cycle policies, all that is respected in the way the data is returned from the user search.

So now we’ve gone from naked GPT in someone’s browser to an enterprise appropriate experience, which you can use in a hospital and will respect PHI, PII, personal information, and even do things like inline DLP. So if I do a search and say, show me Chad’s employment letter, find Ford’s payslip, find Chris’s performance improvement plan, it would block me from doing that because there’s inline data data loss prevention. So it will stop me searching for things I shouldn’t be searching for. And if something does pop up because we’re doing it inside the Microsoft tenant, that will show up in in in some of the data reports.

The data security posture management show that last month, the following data was returned to some prompts people wrote. So now we might tweak our sensitivity labels, or we might tweak our data security policies so that we’re getting safer and safer over time as we’re learning from actual experience in the in the tenant.

So that that’s very interesting because in the very early days of Copilot and and the GPT models and and everything out there, you know, security was a big risk. And we saw it in the chat here. Many people, you know, have an issue with trust in and around that data and in and around that information. So seeing that that has all been been worked out and can now layer multiple models into the mix. And and the great thing when you shared your screen was I’m seeing Opus four point seven on on the list there. That’s the latest and greatest version.

There’s no delay in being able to utilize those models. One month.

Yeah. One month. So Microsoft SLAs Okay. Will follow the latest models by one month to do their testing, validation, integrated into the into the platform. But, you know, in in the grand scheme of things, that’s nothing. So to your point, you will have access to the latest model a month after it comes out.

Yeah. So I think, you know, one of the key things as we move this conversation forward is is, again, going back to you’ve addressed some of the concerns, and you’ve addressed that that big one that I started the conversation with, like, that kinda slap in the face of, you know, Copilot’s no good. Now we understand the premise behind it, what Microsoft is trying to do. They’re not in the model race whatsoever.

They wanna use best of breed to achieve the best possible outcomes for the customers in keeping with a very specific security guardrails on that customer’s data. Now this is applied across the board. This is not something that is necessarily a, you know, an enterprise level situation, but it’s also goes down into the SMB space. Those customers with twenty five and fifty employees have the same security availability as as the enterprise does.

Correct?

Correct. One hundred percent. So they’ve got access to the same tools. They might not go as far in defining data classification policies and sensitivity labeling, but they could.

And they have all those capabilities and all the all the controls are there, and that’s where we, as partners, help those organizations turn on those capabilities and do so in in a way that’s appropriate for their size and their their age and stage. And so it’s all about making sure that the AI experience is appropriately set up for the kind of data they’re handling. Like, if you’re running a if you’re in a business that doesn’t have very heavy security requirements, you might have a different experience Sure. Compared to if you’re in a regulated industry.

Yes. And you’re and you you and you have to pass audit and and have some compliance reporting. You’re gonna need to set up your your AI capability quite differently.

But one of the key things here is the grounding and the training. So if you use ChatGPT, naked ChatGPT in the browser, every time you do something, you are training the GPT model. Okay? It’s learning from you.

Every time you upload a document to get it reformatted or or whatever, you’re now providing your company data to the model. It’s gone. Your data’s gone. It’s out in a public large language model, and you’re training somebody else’s model.

When you do it inside Copilot, Microsoft is not using your data for training anyone’s model. That’s one of the principles of their AI mod their AI practice is that all the work happening inside Copilot is staying inside Copilot. You are not training somebody else’s model, and that’s huge.

That that is huge, and that’s key. Just touching back on the services that you offer, will you do any of the security, the the labeling, anything like that to help set up a customer in a, like, a pro serve or ongoing support?

One hundred percent. That’s that’s what we call work stream number two. We got five work streams in our go to market model, and this is all part of the monthly recurring revenue service. The the the second the the the second work stream is the security, which is huge.

And and and for those of you who are technical or semi technical, you know, what that typically entails is firstly understanding where the enterprise data lives. Some of that might be in legacy on prem file share. Some of that might be in SharePoint online. Some in OneDrive.

Some in third party. Need to find where the data is, understand the nature of the data, then probably figure out what the what the life cycle policies should be, apply some controls, some some labeling, and get the data to a point where we’re able to allow Copilot to access most of the data in real time and surface that data in a secure way without creating embarrassment or breach privacy breaches or security breaches for the organization. Now that’s a journey. Right?

Yeah. Because nobody’s data is clean. It’s all a mess. It’s all a mess. And so we have to assume that there’s going to be a a journey to get the data in the right place, clean it up.

But, of course, we we can’t wait for all that to be done to start using AI because that that cat’s out of the bag. So we’re doing this. You know, we’re we’re fixing the plane in flight and cleaning up the data in real time as we’re as we’re going on going through the rollout.

So what I have found, Denis, and and your team, Chris and Ford, is that customers that are coming to our partners and they’re saying, I want AI. And and they and that’s where the that’s where it stops. It’s a statement. I want AI. And the the pivot for a a lot of our partners is, well, what kind of AI do you want?

One of the talk tracks that that I like to talk about is most likely, you’re already paying for a version of AI because ninety percent of our customers are leveraging Microsoft. So you already have some kind of version of Copilot built in, and it’s really leveraging, again, suppliers like yourselves, specifically, that will come in and help data cleanliness, data security, and set that up in in a manner that makes sense. And not only that, but we’ll also train the the customers on how to utilize it properly. And I think that’s one of the key factors there going back to that MRR conversation as well is this is something that if if the customer is willing to start small, maybe an HR department, maybe something along those lines, where there are high regulatory aspects to the information that will be be shared and searched and all that type of thing and get a win on just that department or just that team and then be able to have that team be the cheerleader for rolling it out to the next department and the next department and the next department.

This is not a go into the company and roll it out for everyone type of of mindset. Correct. Correct?

Yeah. Except in if there are a few examples where people are doing that, but that’s that’s generally not the not not not the recommended way to do it.

The the the other key consideration here is the number of different tools organizations will deploy.

You mentioned that, you know, most organizations already have they probably got the free version of of Copilot Correct.

The chat version. I bet you, they’re probably also subscribing to ChatGPT. There are some people subscribing to Clone. There are some people subscribing to maybe some of the Adobe tools, the Canva tools. Yep. So I’ve got a crazy bold prediction in this space. If you look at what’s happening in enterprises today with security, and you know this well, most enterprises have ended up with something around fifty two secondurity tools.

Fifty two, five two.

Because we’ve been under attack, and we you know, we’re we’re we’re we’re buying out of fear. Over the last ten or twelve years, we bought all these security tools to protect ourselves.

My bold prediction is that enterprises will end up with about fifty two AI tools over the next ten years. It’s driven by a different human emotion. It’s fear of missing out.

Not fear of being attacked and breached. It’s fear of missing out. Yep. People would buy all these tools, and some of them will be used wisely and some will not.

And and that’s a great statement. And and I I will reference myself, specifically my son who is sixteen, has seven or eight different models on his cell phone on his cell phone. And and the conversation revolves around, oh, well, I use this one for this specific task, and I use this one for that task, and I use this for, you know, the the physics problems that I have to do and and all those types of things. It’s interesting how certain models are developing in and around certain areas. Like, there’s a going on in the chat here.

Yeah.

You know, Ford and and Chris are doing a great job with Arvin’s question in and around Claude code and and aspects around that. Claude’s known for its coding capabilities. So I agree wholeheartedly with you, Denis, that this is something that is, a, not going away, and, b, a lot of this information will be utilized specifically for different tasks. And Yeah.

The FOMO aspect is very, very real. If if I don’t have Perplexity to do web searching, then why aren’t you using perplexity? Don’t tell me you still use Google. Like, you know, that those aspects of things are are absolutely true in this.

So let let’s go, you know, to to the next aspect of things here. You wanna take us through the journey of of what’s going on and and how you as Mobile Mentor kind of approach this? Yep. I I know you have kind of a step by step process.

We do. And and I’ll tell you the story of of the of the mistakes we made first because that’s gonna be really helpful to understand where we landed. So when we started embracing AI about three years ago, we ran competitions internally to come up with the best ideas. And we put some prizes, and the first one was a thousand dollars and then, I think, five thousand dollars.

And and, you know, as an organization, we’re just under a hundred people. We’re knowledge workers. We’re we’re in front of our computers all day every day. And so we thought I thought, as the CEO, we really need to lean into this. We need to embrace this. We need to encourage people to figure this out, and we were very, very forward leaning and enthusiastic about this.

And we started rolling out Copied paid paid licenses to about seventy percent of the organization, and then we kind of ran aground, and we realized we had got some things wrong.

And what I now call our five pain points were we had not defined the use cases where AI should have been able to make a difference. We just hadn’t defined the scenarios. Secondly, we hadn’t done our groundwork on data security. We didn’t we hadn’t protected our data like we’ve been talking about for the last few minutes. We hadn’t done that work. We didn’t train people properly, and I felt I was the biggest ******* in the company because I just couldn’t use Copilot properly and Excel and PowerPoint and Word and different things. So we just didn’t have the skills.

We’d had no idea where to start with agents, building autonomous agents that can do work for us. Just didn’t have a clue where to start. And then lastly, we had no governance and no visibility of any return on investment. So when the board said, like, you’re spending all this money, you’re in in incentivizing people, you’re paying for all these licenses, How’s that working out?

I had zero data to be able to say, it’s great. We’re saving time and money, or it’s a complete waste of time. We’re turning it all off next month. I had no visibility whatsoever.

So those became the five pain points, defining the use cases, securing the data, training people to use Copilot correctly, building agents, and some kind of ROI and governance model.

From there, we built our Copilot mentoring program, which has been a runaway success for us, but it came out of pain. And so if we can share that one slide, please, that we sent earlier. Thank you. So the Copilot mentoring program, it’s an MRR service.

So it is a year long program award. It can be multiple years for large enterprises. And it starts off with the business analysis. Going into the business, and I go back to the hospital example, this is that’s so top of mind.

Understanding a day in the life of each of the different clinical users or the revenue cycle people or the administrators or the salespeople or the insurance claims people to understand where AI can move the needle for that particular role and documenting it and saying, here’s the hypothesis. If we give this person AI tooling, they will work better, faster, produce more outputs, whatever. Second program work is the data security, scanning the data, see where is this, what’s going on with the data, and what’s the degree of our data oversharing risk because the LLM search will find everything.

Every offer letter, every payslip, every performance improvement plan, every Social Security number, every credit card that ever existed, way back to eternity, AI will find it. The third is training people how to use, like I showed you here, the researcher, the coworker, the analyst capability, how to restructure documents, how to build PowerPoint decks, how to create Excel formulas, how to reverse engineer Excel formulas. You know, someone gives you a spreadsheet and there’s a ridiculously long formula, and you wanna know what the hell is that doing? You can use Copilot to figure out, you know, what’s the formula.

And then agents, building agents in the right way, and I’m gonna come back to that in a moment because agents is gonna become a mushroom cloud inside organizations. Literally, it’s just gonna go up ballistic on this. There’s some guardrails around agents. And then the last part is the ROI model and taking actual usage data out of Viva to see where people are using Copilot, you know, as a PowerPoint, Word, Excel, blah blah blah, and and getting some proxy for usage and time saving and then building a Power BI report to show the return on investment.

So then we close the loop, and we can say, for the use case we defined for that group of clinical users, we can now see a report showing that, yes, they are getting great value, and we should pour gasoline on that fire and give Copilot to more and more people, or it’s not working. We were wrong. The hypothesis failed. Those people aren’t really using it. Pull the licenses back from them and give some other teams a chance.

And being able to make an informed decision based on actual usage that we could see in the tenant.

And and this is key, and I’m seeing the chat kind of bubble up here again about this. So, you know, again, your model is is and and accurately named is to become a mentor for these customers to be able to help them through the process, from beginning to end. And one of the key questions that I wanted to highlight was, Chris Phillips asked is so if if a company is looking to hire people for AI, you can offer this as a service versus taking on the if that company taking on new hires to do this work.

Correct. So, essentially, it’s it’s it’s, you know, teach a man to fish rather than giving him a fish. So we’re not gonna do staff augmentation and put put people into their organization. We’re gonna take them through this, and we’re gonna show them over time, here’s how you identify your use cases and and document them.

Here’s how you secure your data, and then this becomes a managed service over time, which I’ll come back to because that’s super interesting from a a revenue perspective. And here’s how we do the rollout and the training per use case, per department, whatever. Here’s how we build agents in the right framework with the right guardrails, and each of those becomes a software product that needs ongoing management and maintenance. And here’s how we measure the ROI and take the conversation up a level for the exec team or the board to be able to show what’s working, what’s not, and how we make decisions on where we apply more investment.

Sure. And I think that just answered Keith. Keith just posted here, are you using recommend are you using or recommending specific KPIs or benchmarks to assess areas for improvement in identifying an ROI for your services?

Yes. And it will be different for each use case. Yep.

So, like or do you wanna give an example of maybe one of the use cases where we had a very clear metric around people producing more or or Yeah.

So you can hear me?

Yep. Yes. We can.

So one of the hospitals we’re working with, they have some, like, preadmission screenings, and they’re having issues with those getting denied. They basically have to meet some requirements, some some legal requirements. So this was a great AI use case. They had basically defined a rubric against these government requirements and are able to upload these documents, and it comes back with a score, gives you a breakdown of each section, say, hey.

This is likely not to get approved, gives you a reason why. And, you know, it’s easy to say, okay. Before deploying this tool, you know, we had ten percent of these documents getting denied. After, you know, it’s it’s less than one percent.

Awesome. And so that so does that percentage become the KPI that they want to measure and report on?

Correct.

Okay. That’s crystal clear. Because I can go straight to the exec the exec team and say, here’s what’s different.

Yeah. There’s another question that popped up on the q and a side, not in the chat here, but it’s from Aaron. And he said, since Claude’s coworker is in preview with, quote, unquote, no logging, how do you sandbox it and gain visibility into what’s happening within Claude?

By doing it inside the Copilot framework. So, again, Copilot is the UI for AI to happen responsibly inside your tenant. So all the things that happen in co work, and it’s a it’s going to be a workflow, all that will be based on your data and Teams, Outlook, OneDrive, all your applications stay inside your tenant and follow all your security policies, your data classification policies.

It it’s all staying inside your environment.

That’s great. So and I know this is gonna generate a lot of questions. So we’re we’re headed into the last ten minutes here, and I wanted to have you talk us through the whole process as far as pricing and what does this look like and the MRR aspect of things. But first, Arvin’s been been talkative in the chat here.

So one of the things is, can you support enterprise customers in North America, Western Europe, ANZ? This is all gonna go hand in hand here. But what coverage area does MobileMentor offer specifically?

And then we can get into the pricing and breakdown and how this looks.

Sure.

Yes. We can support customers pretty much anywhere in in in the western world that are using Copilot and following the the Microsoft licensing model. We have operations all across North America, Australia, New Zealand, and then a few team members in different countries in Europe, Middle East, Latin America, different places. So so yes to the g geographic question, and we’re Microsoft global partner of the year for the work we’ve been doing with Microsoft.

We’ve either won or been a winner in the Microsoft partner ecosystem every year for the last five years, which means I’m not saying that to brag. I’m saying that as a way of telling you, we can get funding from Microsoft. That’s a key factor. Yeah.

Yes. And that’s super valuable. So we can get access to their wallet in a couple of ways.

And so sometimes, you know, the a customer will want to do something, and they’re committing to to making a license commitment, and Microsoft will put some funds give some funds to the partner to help kick start that work, which helps enormously.

I see a question here as well. This is not billed by the seat. It’s billed it’s like an enterprise fee. It it typically starts at, I think, six thousand seven hundred and fifty per month.

And so as a per year, it’s about eighty k. And it’s a team of people. So we have a used one specialist who’s doing the use case work, and and the person who leads that team is the former head of clinical informatics from a large health care provider. She worked there for seventeen years.

So she she’s a nurse, and she knows how to find her way around hospitals, for example, and and and and find the use cases because she knows how the data flows between all the clinical applications. She’s able to find how, you know, AI needs to move. The data security, we’ve got data security architects who’ve been working in Microsoft environment for years. The training, you know, we started out as a training organization twenty two years ago.

And Ford is a key player in our development. He’s an AI developer, and then we’ve got a bunch of capabilities there. And and the key point I wanna make around that is it’s not just about developing agents.

Where the puck is moving is it’s about managing agents.

And Microsoft have a new product coming out, I think, on the first of June called agent three six five. That’s where all the money, all the investment, all the funding is going to go on the next financial year. And agent three six five becomes a control plane for all the agents that are going to pop up across the enterprise. So you’re all used to Entra as the control plane for identity. Intune is the control plane for devices.

SharePoint is the control plane for all the the data. Purview is the control plane for securing the data. Agent three six five becomes a control plane for agents. Every agent will have an identity, will have APIs that will call, will have permissions, access privileges, will have a software life cycle and version management.

All of that gets managed in agents three six five as the control plane. That becomes super important, and that’s the managed service opportunity for all of us. Yep. Two opportunities here for managed services.

One is the ongoing data security posture management, DSPM, managing how the data is flowing, what data is showing up in response to prompts. Second, agent management.

Organizations that have tens and hundreds and thousands and tens of thousands of these autonomous agents, every one of them is a software product with a life cycle and a risk profile.

So the managed service here is is just phenomenally exciting.

And so we’re selling this as a as a as a subscription service, and it will evolve as a key thing. It will evolve into a managed service or a co managed service in year two or three as AI gets deployed at scale.

And and I think that’s, again, reiterating what we’ve talked about already is going back to this and saying, again, small wins are the key factor. So being able to implement this in a specific department within an organization first, get it set up and trained and utilizing it, mentoring or training those employees to utilize it correctly and how they can best get the information that they typically require and become more productive, then using that win as an example to continue to expand it over time throughout the whole organization so that it’s tailored properly. Am I am I stating that correctly?

Yes. And I just wanna pick up on one point that’s I’ve seen the chat. I’m just glancing at some of the chat, but somebody asked about fine tuning a model. And and so far, we’ve only talked about Copilot, which is really the canned version of AI for organizations that wanna go further and build product capability for their for their organization.

That’ll happen in the Azure environment, which has got a a ton of capability. So if you wanna go into the Azure environment and build your build your AI tooling, you could choose any model you want. Grok, DeepSeek, Google, Gemini. You you choose any model you want, and you can choose all the data tooling with Fabric and and build out with Foundry.

And you’re that’s real plug and play, and you’re choosing, you know, vision recognition or audio recognition or speech recognition capabilities. And you’re plugging all these things together to build your own model in the Microsoft environment within with all the security you would normally have there. So that’s a that’s a different skill set. Yes.

Different skill set. And so we’re moving into that progressively, but it’s a learning curve. Right? So we’re learning.

You’re learning. The customers are learning. It’s an extraordinary time to be in this business because the the rate of change is just mind boggling. But, you know, organizations would probably start by rolling out something like Copilot, get to a certain point, identify where the real value is for their organization, and say and then decide, instead of paying the subscription for this canned version, let’s build our own, and let’s go deeper and customize this.

Exactly. And and and that’s something, especially as you go upmarket, that that’s where all of this kind of comes together and says, hey. I need this tailored to my needs as an organization, not not a it serves everyone and we’ll we’ll do some minor tweaks to help it help me type of thing. So I I agree wholeheartedly on that. One question that that I do wanna continue down the path on on the smaller side of things. So sixty seven fifty is the monthly recurring. Do you offer anything on a smaller scale for, like, a certain number of employees, or is it really that’s the that’s the bottom line for you in in the offering?

We have done smaller engagements. You know, we’ve had smaller organizations who say, we just need some some training or some guidance or we just need help with the data security piece of this, and we’ll we’ll customize it.

Okay.

So with the any one of those five work streams and and do one of those, but but that will generally create a gap then because you’ve done one thing and you haven’t done the other four.

Right. And so this is best practice to kinda treat this as a program that happens over a year, take people on the journey. There’s a weekly cadence. We’re all working together, you know, a weekly cadence of meetings working through the program. But we can do these individually if that’s what the client wants.

And and and I think that’s key. One of the early on questions in the chat was was around so you can step in without an organization having to hire an AI expert or a number of people and do this for them. The cost factor there, as we’re as most of us are aware for an AI an AI engineer or prompt engineer these days is is astronomic. And let alone can you keep them a full year before they’re sucked away by somebody else.

Yeah. So leveraging a a mobile mentor to come in, help help the customer first and foremost, to help them determine what those outcomes are that they need, getting those small wins, growing it, and moving it along is is well worth the investment for that organization. Understanding that the SMB space, you know, may not have the budget for this, but if they’re if they’re looking at this and it’s a thirty to fifty person company, you know, they it may be worthwhile for them to make the investment and then see how how they can really become more productive. Because, again, the ROIs are gonna be there, most definitely as they roll this out.

Absolutely. And especially if those thirty to fifty people are knowledge workers, you know, sitting in front of a computer most of the time. Right. They really have no choice. You’ve got to figure out how to embrace AI because that’s, you know, potentially how you get you you those fifty people then do the work of seventy people.

And Yep. Exactly.

Where the growth comes from. And so, you know, I’m spending, you know, a few grand a month to figure that out and and get more work done, produce more.

That’s that’s super exciting.

So, again, I’m gonna go back here because the the the it kind of brings this all to a to a nice conclusion here as we wrap up in the next couple of minutes is, Keith mentioned, are there discovery questions that an adviser can use to start the customer engagement? And I think that goes to specifically your capability and capacity assessment that you, really kinda drive the initial thing initial conversation forward with. Correct?

Or it could be the Copilot readiness assessment, which is probably gonna be the more appropriate one here. It. Can you put the link in the chat, please, for the CRA, the Copilot readiness assessment?

Yeah. And and and and, Chad, the way that works is that service has a price because there’s a cost for us to deliver it. Sure. When we get a qualified referral through one of the one of your partners, we waive that cost.

So it’s a it’s a real enterprise with real intent to to do AI at scale. We’ll do the co pilot readiness assessment at no charge. And the value of doing that is we get to understand their maturity on each of these five topics. You know, how well have you actually identified your use cases?

What what does your data security look like today? The the training and the change management maturity, where you have agents and integrations and API availability, then the governance model. We score them. We show them how they compare to others, and then we’ll build a plan and say, here’s what you would need to do to do this responsibly, safely, at scale, and be successful.

And, you know, we like to have you, the partner, part of that conversation because now we’re all on the same page. We’ve all seen we’ve all seen the pain. We all see the opportunity.

We’re all in this together, and we can take the customer in the journey, and everyone’s aligned.

Yeah. Yeah. Exactly right. And and I truly appreciate, Denis, you and the team, Chris and Ford, being able to join us today and talk us through this.

I think it’s important that our partners understand how this is changing business, specifically here in the US, but around the world, and how the adoption really needs guidance, for AI across the board. And I love the term the UI for AI. I’m gonna use that. I’m gonna steal that and use that from now on.

It’s fantastic.

This is this is something that I think can be seized on. I know that the chat is blowing up with questions in and around price, and I understand it may be costly for some SMB, but the point was raised in the chat as well.

How many thousands of dollars are being spent on on tokens and and professional memberships in all of these different models? If I have five of them and four of them are pro, you know, that alone is a hundred and twenty to a hundred and fifty bucks a month just for me.

You know? So, again, it’s it’s something where I think the AI readiness assessment, the Copilot readiness assessment that you offer is something that is a huge value for the partners to leverage and begin that conversation.

And I’m seeing a lot of questions related to customer profile. I guess we’ve designed our service and our model around the SMC market rather than the SMB. So the way we think about small, medium corporations is probably a hundred plus employees up to a few hundred. Most of our services, most of our pricing is targeting that market. Yes. We can go down, and we can we can deconstruct services and reduce services, of course, and we go up into the enterprise space.

But but but the spot we aim for is probably the one hundred to three hundred seat organization. That is our ICP.

Okay. Great. Well, again, I thank you very, very much for joining us today. It’s been a a great conversation.

Really, truly appreciate the guidance for for myself as well as for our partners here today. I look forward to future engagements. You guys are are leaning in very heavily in in the channel, and we really truly appreciate that. So, again, I wanna thank you.

And with that, I’m gonna turn it back over to Cassandra to wrap things up.