Apply AI to pricing and billing workflows with Zuora MCP

Welcome everyone, and thanks for joining us today. I’m Jesse Etang, Technical Product Marketing Manager here at Zuora. I’m excited to talk about one of the biggest shifts we’re seeing in monetization and operations right now, moving from AI as a tool for answers to AI as a tool for execution. Today, we’re- we’ll explore how organizations are applying AI to real billing, revenue, and pricing workflows, not just experimentation with chatbots, but actually accelerate- accelerating operational work across order to cash. We’ll see, uh, examples from customers and how they’re using MCP today, and what they’re looking, um, what they’re looking to learn along the way. Before we dive in,

I want to introduce today’s speakers. Joining me is Sheetal Prakash, Product Manager on the Zuora Platform team and one of the key contributors, um, behind our MCP strategy and product direction. Later in the session, we’ll hear from Paul Farrell, um, who brings extensive monetization and transformation expertise, and Emmanuella Bello from Visma, who will share firsthand experiences using MCP in a complex enterprise environment. We’re looking forward to, um, making this as practical and real-world as possible. So here’s what we’ll cover today. We’ll start with why operational AI matters and the broader shift we’re seeing in the market. We’ll introduce MCP and explain

how it connects AI assistants with business operations. We’ll explore emerging monetization and pricing use cases. We’ll hear from our guest speakers. And then we’ll wrap it up with key takeaways and Q&A. The goal is for everyone to leave with a clear understanding of where AI can drive real operational value today. I want to frame the broader shift we are seeing across order-to-cash operations. Historically, many teams have operated in what we are calling the old world way. Workflow is static, API development is manual, data is spread across billing, revenue, and integration systems, and teams often depend heavily on IT or admin to make even routine changes. That model creates friction.

It slows down changes, delays operational improvements, and makes it harder for business and technical teams to move at the same pace. With MCP, the goal is to change how work gets done. Instead of starting with a ticket, a script, or a manual investigations, team can start with a plain language request. MCP can help translate that request into action across Zuora data, workflow, and business processes. The important point here is not just that AI can answer questions, it’s that AI can help execute operational work, building workflows, accessing the right data, reducing the handoffs that typically slow teams down. Next, let’s define what Zuora MCP actually is and the type of work it

supports. At its core, Zuora MCP is the intelligent execution layer that allows AI assistant like ChatGPT and Clods of the world to work securely with ta- Zuora data and processes. The important distinction is that MCP is not just a chat interface on top of Zuora. It gives AI assistants a structured way to understand what actions are available, retrieve the right information, and help complete operational tasks in a controlled way. Everyone gets excited about the first box, asking questions in plain language. That’s the flashy part. But the box that actually changes how your team operates is the last one, reusable skills. The first time someone cracks a messy

billing investigation, that becomes a skill anyone can run next week. The value compounds. Month three looks nothing like month one. For customers, that means faster time from intent to execution across order to revenue. Now that we have defined what MCP enables, let’s take a look at the early results customers and teams are already seeing. One of the most exciting aspects of MCP is the speed at which value is showing up. We are seeing debugging and issue resolution shrink from roughly an hour to under a minute. Customers are accelerating development and integration work that previously required weeks of work. MCP already exposes hundreds

of operations across billing, revenue, and integrations. And it’s designed for enterprise-scale environments, including lots of subscriber volumes and complex workflows. The broader theme here is that AI is becoming an execution layer, not just an analysis layer. The bottom row sums it all up. Hours to seconds, less manual effort, faster experimentation. Rather than walking through each box one by one, I want to use this slide to kind of highlight the common patterns we are seeing across customers. The first pattern is around acceleration. Customers are using MCP to move faster on development, onboarding, and workflow creation. These are areas

where team often lose time because the work requires context across multiple systems. The second pattern is access to insights. Natural language querying is important because many operational questions sit with business users, but that data often requires technical knowledge to retrieve it. MCP helps reduce that dependency by making Zuora data easier to interrogate. The third pattern is experimentation. Pricing, packaging, and usage models are all becoming more dynamic, but testing those changes can be complex. MCP can help teams evaluate scenarios better before making production changes. So top row says, “Stop being blocked.” Bottom row says, “Do things that weren’t even possible before.” The bigger message here is that MCP

is not limited to a single department or a single business flow. It creates a more action-oriented layer across Zora teams, helping technical teams and business teams work from the same operational context. With that foundation, we’ll move into customer examples and demos to show how these patterns come to life in real business flows. Okay. And now I’m gonna kick it over to our first customer spotlight. I am joined with a Zora consultant out in the wild, Paul Farrell. Paul, how are you? I’m good, Jesse, how are you? Good. Uh, do you wanna tell the audience what it is that you do? Sure. So Jesse, I’ve spent most of my career building monetization, billing, subscription,

and payments platforms for a company like Cisco, Zoom, Matterport, and others. One reason AI technology has been interesting to me is that software monetization is going through the same kind of transition that infrastructure went through with cloud computing. AI agents are starting to become participa- participants in the design process itself. I was given a chance early in the development cycle to evaluate Zora MCP, and that’s why I’m here today. Nice. Okay. So do you wanna tell us what business challenges led your team to explore MCP? Well, you know, the challenges we all face is, how can we do more with less? And how can I make sure that I’m not falling behind? I’ve experimented with, with other MCP servers

and see their value in implementing agentic workflows. The other challenge is that my team is pretty small compared to the kind of structures at other companies, like when I was at Zoom. Nice. Can you describe, um, your token monetization model that you were building? Yeah, sure. You know, I’d looked at the advanced consumption billing features earlier on, but did not have a commercial example at the time that needed implementation. To be honest, the documentation is pretty weak, and working samples are not readily available. I wanted the AI to start simple and add complexity that mirrors how AI companies evolve monetization models like pure usage, committed plus overage, token packs, and

rollover plans. You know, and based on the outcomes of this experiment, I have some more frontier-type monetization models that I want to build in the future. Nice. So how did MCP helps accelerate all of your pricing and monetization experimentation? Well, it helps as the server exposes curated commands to, and sub commands to my LLM client. The reasoning process of these LLMs is to favor functionality exposed via MCP versus trying to call APIs directly and potentially getting hallucinated responses. The server was able to churn away at the requirements and produce some pretty decent results in a few hours. Wow, that’s awesome. So what did MCP actually automate

that would have otherwise required significant manual effort? Well, for the last 20 years, when- whenever you wanted to launch a new pricing model or a subscription offering or usage-based services, you would typically assemble teams of architects, product managers, engineers, and operations specialists to design the solution. The challenge wasn’t usually knowing what we wanted to build, it was figuring out how to translate an idea into a working implementation across multiple systems. By understanding our intent, MCP servers are able to build out these models and iterate to fulfill the request. MCP servers in conjunction with LLMs not only knows the details of the Zora structure, but how users intend to use the billing system to achieve their outcomes.

That’s awesome. So what do you think, um, the opportunities that A- AI and MCP create for future monetization initiatives? Well, as monetization professionals, we should not just be order takers, but strive to be trusted advisors to the business. Love that. A lot of the time, our users don’t know what’s possible to model and operationalize. You know, we too are challenged to keep up with what’s possible, and AI help us to create innovation cycles to discover these monetization opportunities. I’m also interested in this, these new, newer outcome-based billing models. You know, traditionally, usage-based billing models have been modeled on

very tangible outputs like phone calls or gigabi- gigabytes of storage or seats provisioned. Cloud or edge deployed models are increasingly creating what I call semantic billing events. And what I mean by semantic billing events are samples like detection of an impending heart failure on a wearable medical device, detecting motor failures before they happen in a factory, diseases in crops. The applications are endless. Being able to monetize these types of outcomes allow the capture of the value by AI models and not just raw AI token counts. Wow, sounds really interesting. Do you wanna show the audience all the things that, um, you worked on? Do you

care to walk us through a demo? Yeah, sure. Stop sharing my screen. Okay. Before we dive into the demo, I want to introduce what I call the AI development loop. Although this slide uses Codex as the example, you could substitute Claude, Copilot, or whatever AI assistant you prefer. The important idea isn’t the model, the important idea is the loop. Historically, software projects followed a fairly linear process. Requirements were written, handed to developers, code created, tested, deployed, and eventually reviewed by the business. What we’re starting to see is a much tighter feedback cycle. A user decides objective in natural language, the AI proposes a plan,

the AI executes against tools and systems through MCP, and results are returned. The user reviews, refines, and steers the next iteration. Then the cycle repeats. Instead of humans spending most of their time creating artifacts, they spend more time guiding outcomes. This feels similar to the transition from physical infrastructure to cloud infrastructure. We move from manually configuring servers to defining intent and letting platforms handle the implementation. MCP extends that concept beyond infrastructure. It allows AI agents to interact directly with business systems such as Zuora. The result is that activities that previously took days or weeks can often be compressed into hours, while still keeping a human in the loop to provide

judgment, governance, and business context. The token demo project I’m about to show was built entirely using this loop. So here was life before AI MCP. Developers being handed a requirement would slave away and do all of the activities in, in, in their silo manually. Slow, unpredictable. The user, again, configuring Zuora, all of these manual activities, and all the time the business user is waiting for the response. And when they do get a response, if they need to do analysis, they’re do- using– doing reports, spreadsheets, waiting for actual data,

and don’t have the information to actually create decisions. Slow decisions, higher cost, missed opportunities. Here’s life after AI MCP. There are three ways to view this project: the developer experience, the Zuora user experience, and the business user experience. What’s interesting is that’s the exact pr- same project viewed through three different lenses. From the developer experience perspective, the focus is on how the solution gets built. The focus is on how the solution gets built. This includes the prompts, the planning process, MCP interactions, artifact generation, code, scripts, workflows, and the iterative loop between the user and the AI agent.

This is where we look at how an idea becomes an executable implementation. From the Zuora user experience perspective, we shift our attention to what actually gets created inside the platform. Products, rate plans, rate plan charges, subscriptions, usage records, invoices, bill runs, and revenue events. This is the operational view that Zuora administrators, billing teams, and revenue operation teams care about. Finally, from the business experience perspective, we step back and look at the commercial outcome. What monetized models were evaluated? How do token packs compare to recurring commits? What happens when we introduce rollover balances? Which model generates the most predictable revenue? This is the perspective that product managers, finance leaders, pricing teams, and executives ultimately

care about. What makes AI-assisted development particularly interesting is that a single conversation can now span all three perspectives. Traditionally, these were separate activities performed by different teams over days or weeks. An architect would define requirements, developers would build solutions, administrators would configure systems, analysts would evaluate results, business stakeholders would review reports. In this project, these activities became part of the continuous feedback loop. We’ll start with the developer experience, watch the project being created, then we’ll move into the Zuora user experience to see what was actually built. Finally, we’ll examine the business experience to understand the commercial insights the project produced. By the end, you’ll see not just a token monetization

demo, but a new way of thinking about how monetization systems themselves can be designed, implemented, and analyzed with AI as an active participant in the process. The goal was to evaluate several AI token monetization models, including pay-as-you-go, commit plus overage, token packs, and rollover wallets. Historically, this kind of analysis could take weeks. The objective was to determine how much we could accelerate using AI and MCP. This is the environment where all the work happens. On the left are the project files. On the right is the AI agent. The conversation isn’t just producing text, it’s producing files, code, test data reports,

and Zuora artifacts. So here we are. We have a requirement in the middle. We’ve specified all of the different models that we want to produce. And you’ll notice that in the file view, the only document is the requirement. So over in the AI agent view, I’m telling it, the AI and MCP, to actually implement the requirement in the open document. But for the interest of this project here, I will also want to record were the inferences made? Is it just Codex or is it using Zuora MCP?

And through the magic of modern television, this is sped up significantly. It’s actually going through all of the requirements, coming up with the plan to implement, and it has decided upon an implementation approach. And we’re seeing where all the inference is happening, whether it’s Zuora, MCP, or the API directly. Also, if… the AI will actually look at other projects that you have locally on your machine I’ve told it not to do that this time because I want it to be a complete clean build. And, uh, so it’s completed the planning process,

and it’s come up with all these requirement- these markup documents which are further instructions to the AI to build out the project. It’s also created, uh, the products on the server, and it’s created all of the invoices and, uh, usage files. So here we’re seeing a summary of what the AI produced. Here is the catalog of the products that the AI created, and here’s the actual view within Zuora of what was actually produced. Let’s go look at the product with the rollover capability.

So this is the product of product catalog view in Zuora. And traditionally, this is like, uh, trying to defuse a bomb because you select the wrong option and, and you’ll get entirely unpredicted results. Hopefully our AI will know what to do correctly. So here’s an ex- an executive summary as well of, of, uh, what, what the system did in terms of the subscriptions that it created. Also, it has come up with a basic analysis of what plan it thinks is best. But you can give… You can guide it and provide more information

in terms of how you expect the analysis to be done. Here’s something interesting. I asked the system to record what failures or what issues that it encountered in the implementation. These are the kinds of things that are, that can waste days for a developer if they get stuck. AI and MCP was able to figure it out automatically. Let’s look at an invoice. This is just a standard invoice template, but we see the token top, top-up charges and the drawdown on the usage. Looks good. So

for the rollover scenario, I looked and saw that the test data that it came up with didn’t actually generate rollover data. So I issued a further prompt to actually create an, uh, an extended scenario that would force this particular condition. So per this summary here, it generated more usage and has actually, uh, validated that rollover events happen. This is a complicated scenario because it’s a challenge to show the rollover data on invoices. So AI will be able to help in terms of building this out if we decide to implement this approach.

So here’s further analysis in terms of the rollover. It has provided a lot of information here so the business user and everyone in the project can get more detail about, uh, how to evaluate this model. I also asked the system to record when NCP was used. This was entirely for my purposes and is a good debugging tool for developers to actually see what’s going on under the hood. And here are some of the actual commands. This is again more for my own edification.

Here are some outputs that the business user can look at. It actually shows the simulation of the data that it created for each of the business models in each of the different months for a particular persona, and the nodes associated with it. This approach could actually be extended where the business user could provide additional information in terms of what usage profiles are going to be happening from the different personas, and it could automatically go through the system again and generate all of the outputs from invoices to revenue. Here’s

a catalog of all of the subscriptions, accounts, and invoices that were gen… This is a different view for the business user to see how the revenue profiles for each of the different business models vary over time. It’s kind of hard to see because AI gave the same color to each of the different models. Well, I did ask AI to produce a different color for each model. Oh, looks like we’re not the only ones looking to sell AI tokens. Hi, Emma. [laughs] Hi. How are you?

Doing great, Jessie. Uh, really excited to be here. Thank you very much for, uh, for, uh, this initiative. This was actually one of m- uh, the topic I’ve been waiting for a while. [laughs] Awesome. Well, thank you for joining us. We really appreciate it. Emma comes all the way from Visma. She is one of the solution architects running, um, their Zuora Um, accounts over there. So Emma, very quickly, high level, please introduce yourself to the audience. Tell us a little bit about Biz- Visma and what it is you do there. Of course. Uh, as, uh, Jessie said, um, I’m, uh, Emmanuella. I’m a solution architect in Visma, and I’ve been working

here for five years. I sit in the business technology team, and my role is, uh, to help, uh, business units onboard into Zuora and integrating their ecosystems like CRMs, ERP, and, uh, various third-party solutions. Um, basically any time a business unit want to change, extend something to connect, uh, their billing systems, uh, to another tool, uh, that’s where, uh, I came in. Yeah. How many business us- units, um, does V- Visma have? Uh, now, uh, to be honest, uh, with the splitting part, um, I think… One

moment to, to be sure. With the sandbox or, uh… Uh, yes, business units. Yes. Uh, sorry. Yes, we have sandbox, and we have productions. Um, 30, yes. 30 business units. Yes. Okay, so a pretty complex system that you have at- Yes… Visma. And they, they having all the sandboxes and, uh, production as well. Wow. So that’s probably, like, over 60 different tenants you have to work through- Exactly… to make sure that they- Yes, and for some o- some of them we have development as well. [laughs] So- Wow. [laughs] Yes. [laughs] Oh, my goodness. Okay, so very complex.

So- Yes… what led you to use the MCP, the Zuora MCP? Ooh, yes. Um, um, I want to mention here first that, uh, I’m, uh, m- leading a, uh, this part of, uh, testing new features in, um, in, uh, Visma, new features released, uh, by, uh, by Zuora. Uh, so everything what was released, uh, Zuora AI, Zuora Help, Zo- uh, Zuora Copilot, everything, uh, what, uh, Zuora released, uh, I tested before. So basically this was the, uh, one, uh, like other features released, uh, by you. But

I said, uh, “Hey, uh, let’s, let’s try. Let’s try to see what, uh, what brings this…” I wanted to understand exactly how it’s working and not just read it about it. We, uh, were already dealing with multi-entity complexity, and, uh, yes, uh, so, uh, connecting, uh, directly Claude to Zuora, uh, MCP f- feel- felt this, uh, like an opportunity to, to test, uh, multi-entity, um, uh, connections. Yes. Yeah. Yeah. Yeah. So what have the results been since your use of the Zuora MCP? Ooh, yes. So basically, um, we, first we integrated this, uh, MCP,

like, um, just, uh, doing stuff, uh, create subscriptions, um, uh, create accounts, uh, so, like, Zuora operations stuff. But, uh, after we, um, we, we see that MCP is capable to do more things, to concatenate reporting stuff. For example, um, we extract reports, uh, through MCP, which, uh, if we go directly in Zuora, Zuora is not capable to, to, uh, give us, so a complex, uh, report. So basically it’s

s- it’s doing stuff, um, um, um, from, uh, concatenating reports. Uh, it’s, it’s doing stuff very, very, uh, complex. So it’s, uh, it’s not like, uh, okay, I’m not, uh… I don’t want to go and log in in Zuora. Uh, it’s, it’s a complex, uh, uh, um, thinking behind how to say. Yeah. Yes. Yeah. So, uh, yes, and after this we, we saw that, um, creating a, creating a skill help, uh, help us, uh, doing things, um, more easier.

Uh, we create a lot of, uh, skills like, um, um, converting a Zuora Word template in HTML. And here I want to bring up the fact, uh, we, we, we spend, uh, days, uh, in the past, uh, just to map the right fields in, uh, in Zuora, uh, document template. Uh, understanding the syntax behind and, uh, validating the output, yes, was a nightmare if you ask me. And now we are, uh, we are, uh, spending hours just, uh, just, uh, converting this in, uh, HTML. Wow. Yes, it’s, it’s amazing, yes. And, uh, not to mention we, we done

some, um, uh, skill, uh, like Zuora operations. So we… You don’t need to, to log in Zuora to create an account, to create a subscription, and to extract information from a subscription to understand the complexity of, um, of a subscription, uh, from, uh, from MCP, uh, without any connection, uh, any connection with the UI. Yeah. It’s, it’s amazing. Yes. Yeah. That’s, that’s really awesome. Yes. So just for the audience, what is a skill? Yes. A skill basically is, it’s turning, uh, all Zuora logic. Um,

yes, all, all Zuora logic, um, the integration patterns, the edge cases into something the whole team can, can understand- Mm-hmm… without being, uh, uh, needed to, to have, uh, um, high level of expertise in Zuora. Yeah. So basically we, we build these skills, um, as I said, like workflow and HTML conversion. Um, instead coming, uh, uh, somebody to, to, “Hey, help me with this, uh, with this thing,” just it’s just logging in, uh, MCP and, uh, um, accessing that skill. “Hey,

let me do this with, with Zuora skill, uh, op- Zuora operations.” And yes, it’s, it’s, um, converting a high level of n- knowledge without, uh, uh, going into do- in- into UI. Yes. Yeah. Okay. It’s very nice because- That’s awesome… it gave the possibility to extract, uh, a lot of informations with this. Yeah. Yeah. So a skill is a way to take complex logic. Exactly. It also adds, um, edge cases, it also adds different patterns that you and your team- Yeah… might have talked through, turns it into a document- Mm-hmm… stores it in your

LLM. In this case, it is, um, Claude, and then it allows you to reuse it every single time you go to do that specific use case. Is that correct? Hello? Okay, you’re back. Yes. Yeah. But- I was just summarizing what a skill was for the audience. We lost you there for a second, but all good. So a skill is essentially a, um, I don’t wanna say it’s like a, it’s a Zor- it, it, it holds Zuora logic.

It takes into account edge cases, it takes into account patterns- Okay… integrations that you and your team have identified. It stores it into a document, and it allows your team to reuse it every single time it goes to do that specific use case. Exactly. So it saves you so much time from iteration. It saves you that knowledge barrier, right? So you might not need to know too much about Zuora anymore to be able to get your job done. It’s really fantastic. So do you mind showing us- Yes… um, how your team is using the skill in real time?

And then I have one more question for you while you’re pulling up, um, your Claude. What’s next for Visma and the MCP? Like, what else are you going to try to do with it? Uh, yes. Good, good question. [laughs] The next is, uh, yes, we are now testing, uh, the last version released by you, and it’s working, uh, nice, pretty nice. Um, the next thing, uh, we are doing all the, we are, uh, searching and, uh, checking all the legal stuff, the compliance, uh, to go in, uh, production with, uh, Zuora MCP. Mm-hmm. We, we, the s-

the business units are, uh, requesting, uh, this, this thing and, um, I think it’s a matter of, uh, weeks that we will release in, uh, if everything is going smoothly on the legal part, uh, it’s a matter of weeks that we are in production. Wonderful. Yeah. Yeah. Sounds good. Okay. Well, let’s dig into your skill making factory. [laughs] Yes. I’m sharing my screen. Okay, let me stop share. Okay.

Good. Do you see my screen? Yes. Perfect. So, uh, as I mentioned before, um, we are using, uh, Claude. Claude. So, uh, MCP, it’s installed on top of, um, MCP. Uh, on top of Claude, sorry. [laughs] And, uh, yes, uh, I can show you here the skills, what skills we have built. Yes, this is the last one, uh, but it’s not, uh, prepared yet. N- not ready yet. Um, sales metrics, uh, Zuora order skill, um, Zuora template converter,

what I mentioned you, that is saving, uh, lots of, a lot of time, and Zula- Zuora design, uh, generator. This skill also, it’s, uh, pretty nice because we work with a lot of, uh, Zuora onboardings, and gives us, um, a clear, um, solution blueprint diagram with all the connections which will be, uh, all the tools connected, uh- Which will be connected to, uh, to Xora, and to have a clear understanding of, uh, what will be connected and how. Mm-hmm. Yes. Um, but I want to show now, uh, just a use case, a pretty nice use case, how it’s, uh,

working. I will use, uh, Cowork, and I will give a new task. I want to show you, um, a report, um, which I say, said to you that, mentioned before that, um, we need to, before we concatenate two or three reports to, to see exactly how the revenue, it’s, it’s distributed a- across one, uh, one, um, uh, subscription with multiple versions. So, uh, for… in this example, uh, I will show you exactly on, uh, this subscription,

uh, how, uh, how the revenue schedule, uh, looks like, uh, the MRR on the subscription during the life cycle, a detailed analyze, uh, on the subscription. Um, and yes, uh, the subscription is greater than, uh, two years. I put some condition as well. Mm-hmm. Yes, uh, MRR fluctuations, how… and yes, how the revenue is recognized, uh, versus, uh, unrecognized. And using, um, Xora skill, I will, uh, prompt this just to show you how it’s working and how

it’s bringing up all the things which I mentioned that lose hours just to concatenate and bring us all the information in one place. Mm-hmm. Yes, and, and you can see all the progress. Uh, each minute, uh, you can see exactly what is happening. Yes, uh, one thing what I like, uh, the most on MCP is the fact it’s asking things. When doesn’t know how to act, it’s asking, “Do you want to proceed with this?” So pre-approval part, it’s the best thing, [laughs] the best thing, uh,

from, from this. And, uh, it’s not acting, uh, bef- uh, before approve it. So yes, this, this, it’s, it’s wonderful that it’s happening. As you see, it’s working, uh, very good. It’s,

uh, doing, uh, queries in the background, which you can see. The query objects, uh, can be seen if you expand the window here. Um, and, uh, all this query which are running in MCP can be visible after in, uh, in Xora if you go and log in. This is the best part as well. [laughs] And I also love that it has, like, a little, like, checklist of tasks that I can see- Yes, exactly… and crossing it off. Yeah. Uh- Yeah… I’m saying every day that it’s my personal assistant. [laughs] Yeah. But, uh, it’s, it’s very,

uh, capable to fetching the data across multi- Mm-hmm… uh, multiple objects. This is amazing. As I said, you… if you go in UI, Xora UI, and you want to do such a complex report, uh, you need to, uh, perform two or three reports from multiple ob- objects. Yes. Mm-hmm. It’s not possible from UI.

Yes.

Some more seconds, but it’s, uh, it’s thinking good. It’s running good. Everything, it’s under control. Yes, um, until, um, it’s finished. Uh, yeah, one, one more thing about, um, uh, MCP. Actually, we, we save a lot of time on, uh, doing, uh, um, troubleshooting on the workflows. You can extract a full report with the, um, with the errors from, from, uh, the tenant. So this is amazing as well because what– from… as you know, from, from the

UI, uh, it’s not possible to do this. Yeah. Yeah. So it seems like we’re on the last task. [chuckles] Yes. I love how you can, like, see its personality. Now I have a very clear picture. Let me also check for revenue schedules. [chuckles]

And here we go. [chuckles] Success. [chuckles] Yes. So, uh, as you can see, a full analyzed report here. It’s, it’s, uh, being displayed. So as, uh, key findings, the subscription life-life cycles from, uh, when beginning and, um, when, uh, ended, how many versions for… Since when is this, uh, subscription? Uh, the MRR fluctuations in three phases, phases as well: phase one, phase two, and phase three. Our revenue recognition, as I requested, uh, versus, uh, unrecognized revenue.

And here, uh, are, um, uh, the exactly fully recognized, uh, um, amount and, uh, twenty-eight quarterly service period which, uh, were elapsed. Uh, and, uh, yes, no, uh, unrecognized, uh, revenue on this, uh, subscription. And what I like the most is the fact, uh, MCP has his personal, uh, opinion, like an accountant. [chuckles] And, uh, it’s saying exactly how many, um, uh, unpaid invoices are.

And here it’s exactly the invoice in, uh, in scope, uh, what amount, um, and, uh, for what it’s covered on the– this specific, uh, invoice. Yes, it’s, it’s very nice. The report is amazing. What the… It’s fully detailed So once you get this report, what then happens? You pass it over to your finance partners, or- Y- yes, yes. Uh, we pass exactly to, to the finance team to check it’s, if it’s accurate. Of course, as- Yeah… as an AI, you need to double-check this information before. Yeah. Uh, after that, yes, to check that, uh, it’s accurate.

Uh, yes, and, um, uh, we, we extract, uh, what is needed to, to complete our view on this, uh, subscriptions. But it’s more than necessary- Yeah… at this level. That’s awesome. Well, this was really- Yeah… informative. This was great for me to know how people are using MCP in their day-to-day. Do you have any last words, um, for the audience? Yes. I encourage everyone to use, um, to use MCP, uh, to use AI, AI, uh, Azure AI, and, uh, to try to understand exactly how it’s working, how it c- can be, um, um,

how could be integrated with, uh, their a- actual, uh, ecosystems. Mm-hmm. And yes, it’s, it’s amazing what you can do with, with it. Yeah. Awesome. Well, thank you so much for taking the time out to chat with me today, Emma. I really appreciate it. Thank you very much as well to having, to being here. [laughs] Yeah. Alrighty. Next, until next time. Thank you. Bye. If there’s one message I’d like for everyone to remember from today’s discussion, it’s this: the biggest gains aren’t coming from AI itself. They’re coming from applying AI to real business workflows. We’re seeing faster execution, less manual work, more confident experimentation, and deeper integration of AI

into day-to-day operations. The organizations creating the most value are the ones embedding AI into processes rather than treating it as a standalone tool. As we look ahead, we’re seeing a fundamental transition from manual operations to autonomous operations, from engineering bottlenecks to business-led execution, from static monetization to adaptive monetization platforms. The organizations that embrace this shift will be able to move faster, experiment more confidently, and respond more effectively to changing customer needs. That’s the opportunity that MCP is helping unlock. Thank you for joining us today. We hope that this gave you a practical view

of how operational AI is evolving and how organizations are already applying MCP to real monetization and billing workflows. We’ll now open it up for questions. Feel free to ask about MCP, scales, pricing experimentation, customer use cases, or anything else you’d like to explore