Customer story
Jul 10
1h 0m

How Yext cut integration costs by 60% and delivered faster with fewer tools

Yext delivered 100+ integrations in 3 months. See how they did it and why their old platform was slowing everything down.

Yext’s integration platform created delays and growing costs. It took too long to build and change workflows, and the cost of maintaining multiple tools kept growing. The team needed a faster way to support new systems, scale delivery, and reduce overhead.

In this on-demand session, Tulasi Donthireddy, Senior Director of IT & Business Systems at Yext shares how his team moved over 100 integrations in under 90 days, cut costs by 60%, and made it easier for both developers and business teams to deliver on time.

You’ll also hear from Alexander Wurm, Senior Analyst at Nucleus Research on what most teams miss when evaluating integration cost and what to look for if you're planning to modernize.

What you’ll learn

  • What pushed Yext to replace their integration platform

  • How they handled the migration with a small team and a tight timeline

  • How removing extra tools made delivery and support faster

  • How they enabled faster delivery across technical and business teams

  • Where integration costs hide and how to spot them, according to Nucleus

  • What matters most if you're planning for agent-based automation

Transcript

Hi, everyone. Welcome to today's event. Thank you for joining. I'm Paul Turner at Trey, your host for today's webinar.

Webinar is how to cut integration cost and triple developer productivity.

And more importantly, why you need to do this with all the AI pressures, that we're all facing, coming down the pipe as well. Got some great advice lined up for you today, and I think you actually can find it a really, informative, hour.

So really pleased, to be joined by my my better looking counterparts here, Alex and, and Tulasi.

Alex is a principal analyst with, Nucleus Research, and Tulasi Dunafreddy, is a senior director by IT and business systems at digital presence leader, Yext.

To give you a little background on our speakers, before we get started, so Nucleus Research. Alex Worm leads research coverage of iPaaS, data management, analytics, AI machine learning, and provides guidance to IT leaders and providers. Did I miss anything there there, Alex? Was that was that all your coverage areas?

That's it.

Do you have a good time to sleep? Do you have a good time to sleep?

We get a couple hours in.

So, Alex, for the event here. So Alex is actually gonna take you through the reasons that if you're in IT, or maybe just focused on delivering solutions for the business, the now's the time really to be focused on being more, agile and taking a close look at the the cost of ownership of your existing integration tools.

Because all the geopress we have, right, existing business needs and all the emerging needs to deliver agents and, infusing out into our processes, a lot of a lot of pressure points. And often integrations are the center of that as well. So, yeah, take a look at the tools you have and try to figure out if they're the right way to move forward. So welcome, Alex.

Awesome. Yeah. Thank you for the introduction, Paul.

And then, over at Yext, Talassie is responsible for implementing solution systems integrations, improving business operations, and he leads an applications team responsible for complex business requirements, turning into solutions, as well. And so, so Talase is gonna take a deep dive, into the drivers of change at Yext, some of the challenges they faced, with the growth, their goals for technology transformation, and also they're gonna share some best practices that you can take back with you, after the event. Welcome, Tula.

Thanks for the intro, Paul. Happy to share our story.

Great. And then, before I, hand over to, to Alex, and to Lassie, if you have any questions, please don't be afraid to ask.

It's we'll just post post them just post them during the event. And if if there's a right time, we'll, we'll answer them, either during or we'll answer them at the end or so whatever works.

And, we'll also share some, share some next steps, for you at the end of the event as well. So some some some some some some action to take as well. So with that, I'm gonna hand it over to to Alex. So over to you, Alex.

Alright. Thank you very much. Appreciate the introduction, Paul. As you mentioned, I'm Alex Worm, principal analyst at Nucleus Research, covering the integration and data management markets primarily.

As with any of them, we've had a little bit of mania as all of these different spaces have been reinvented by AI. So, naturally, I'm gonna try to walk you through how this applies to iPaaS and some of the considerations we're looking at in our research.

At Nucleus, my research focuses mainly on delivering credible ROI analyses alongside customer driven insights drawn directly from our engagement with end users.

Now this approach is important to us because it ensures that our recommendations really reflect real world outcomes rather than just the theoretical predictions.

So today, I'm gonna share a few key insights and takeaways from our recent research.

First, we'll look at the iPaaS market, the state of it in twenty twenty five, and how innovations like AI are really changing the scene.

We'll then explore some considerations that I have for investment teams, how they should prioritize their technology decisions, and what features they should look at when evaluating iPaaS solutions specifically.

Finally, I'll set the stage for Yext to illustrate the practical implications of our strategic technology decisions.

First, let me give you a quick introduction to Nucleus. Paul, I'm sorry. Can you just scroll back a few slides there?

Yeah. That's that's fine. Perfect. So Nucleus is an independent technology analyst firm, and we focus mainly on, value messaging and on real consumer outcomes.

We have published over a thousand ROI case studies over the past twenty five years, each validating technology investments with ready metrics that are relevant for those organizations and for financial teams.

To this day, we are still the only registered firm with the National Association of State Boards of Accountancy, providing an extra layer of rigor and transparency to our research that allows financial professionals to trust the analysis that we deliver.

We also deliver advisory services to end users and vendors that build off of this research, following experts in the space.

Now let's, talk about research. There's a challenge to some of the research that we deliver, especially with the enterprise.

Enterprise buyers are really interested in findings that are tailored to their unique circumstances. They want personalization. They want individualization.

And so an ROI model that might fit a high gross SaaS provider rarely maps the same way onto maybe a capital intensive manufacturer, a retailer, or a different type of business.

So importantly, we have to consider things like scale, process maturity, regulatory load, and what competitive pressures an organization faces in order to give a really nicely tailored, piece of research.

So to meet these expectations among enterprises, among organizations, we have two complimentary research tracks that you could see up here.

The first are broad surveys and value matrices.

These help us identify macro trends and present general vendor positioning across two of the most important areas, which are functionality and usability.

These are highly general in any ROI analysis, the solution that is going to impact the most of your users with the most functionality and feature set is going to drive a high return.

So these are good general studies. But how do we feature that personalization and drill into these individual organizations? We do this with ROI case studies and with customer research itself. This allows us to look into line item cost, industry specific measures, productivity impacts for certain roles, and really drill into how your organization could use a technology.

Putting these together lets us address both universal questions as well as situational questions.

The matrix may tell a CIO what platforms excel at usability, whereas an ROI study answers, you know, what savings can my organization realistically book within a time frame.

Combining these two equips decision makers with both perspectives to achieve current results and achieve big picture returns.

Now with that understanding, let's set the scene and review the iPass market.

Five years ago, the median enterprise juggled just over a hundred apps.

Today, that figure tops three hundred.

Every additional marketing automation tool, niche finance module, or AI assistant drops another endpoint into the mix.

Each one admitting data that must be captured, cleansed, and moved in near real time.

The stack is no longer a tidy pyramid of enterprise apps like ERP, CRM, and HR. Now it's more of a constellation with different stars drawn together and piece together.

As you layer AI on top of this, the velocity accelerates further.

Low code model builders using AI let teams and developers spin up new services easier. And all of this innovation transfers into greater traffic and greater application sprawl.

Without a modern iPaaS solution that could actually deal with this scale, you're going to encounter greater challenges.

This is why the market is gravitating specifically towards cloud native iPaaS solutions.

These platforms ingest streams, apply governance, and can scale horizontally the moment a new AI service goes live, so you don't have to rewrite or rearchitect your system.

In short, it gives the organization more agility to adjust to new services as they come live.

Now let's address what many teams have discovered the hard way over time.

The first generation of iPaaS simply hasn't kept pace with the way that we build software today.

Workloads arrive as real time event streams, updates deploy continuously, and AI services expect millisecond responses.

Yet, licensing for many legacy tools is still pegged to connector counts. So every new endpoint increases OPEX.

To plug these gaps and control costs, teams bolt together on point solutions. They add an API gateway here, a low code mapper here. And engineers end up integrating these integration tools just to keep their data moving.

In our research, we've shown that this patchwork alone drives fifteen to twenty nine percent higher total cost of ownership relative to a unified platform. While siphoning sixty to eighty percent of the average IT budget into maintenance instead of innovation.

This hits and hurts the most in agility and productivity.

Every extra integration challenge pulls developers off road map work.

Release cycles slow, customer wait time stretch, and promising ideas stall because the plumbing can't keep up.

When dollars and developer hours disappear into upkeep, integration stops propelling growth, and starts blocking it.

This operational drag means that I team teams often become reluctant gatekeepers rather than partners for the business.

Now switching on to agents, Paul. Thank you.

AI agents have moved from a proof of concept over the past year to production, and this shift exposes two main pressures that are relevant to integration.

The first is API management.

Second, agent orchestration.

AI agents often depend on tool use, which usually means that they have and necessitate clear API documentation, valid secrets, stable endpoints, and reliable connectivity in order to do the workflows that they're assigned.

When any one of those is missing, the agent stalls out, and the user sees an error instead of an answer, completely independent of the model performance.

AI agents as they stand now are also far from fully autonomous, far from agentic AI as we're being pitched.

Most real world tasks currently involve several specialized personas or different roles of these AI agents that work congruently in order to complete a task.

Without a single place to coordinate these agents, tasks slow down and mistakes can creep in.

Modern iPaaS solutions form this foundation.

They keep every credential in one encrypted vault. They apply the same rules to every request, and they record a single log that everyone can trust.

They also are starting to include agent builder systems that allow teams to lay out personas, connect their steps, and run heavy workloads across MCP servers, across emerging a to a protocols, and move data straight between apps without additional middle layer.

The result is just much better organization, observability, and visibility into processes that currently lack all three.

Integration becomes the operating layer that keeps AI innovation fast, safe, and scalable.

In other words, modern integration platforms act as this facilitator for AI rather than just another layer of infrastructure to manage.

So here's a checkpoint question that every architecture review should start with.

Is our iPaaS actually ready for AI?

The iPaaS that wins tomorrow will not be the one with the biggest connector catalog like some might market, but it's gonna be the platform that best supports expanding AI services.

Integration now sits on the critical path for every AI feature.

Agents can only act as fast as they could hop between different systems. And if the integration layer lags, the agent stalls no matter how smart the model is.

We expect two factors to really separate the leaders in this next stage of iPaaS from the lag glaggers.

The first, as we previously mentioned, is real time orchestration across agents, bringing together these different roles.

The platform should broker calls between different agent personas. It should handle retires and route messages based on live context.

The second capability is dedicated monitoring and visibility.

Platform should provide unified tracking and usage analytics showing exactly what each agent is doing across apps. So teams can spot issues, optimize performance, and see when a human is necessary within the workflow.

When these pieces come together, they allow you to go into production in much shorter of a time with new features, especially AI features.

Next slide, please, Paul.

So all of this comes back to a single question. We have a lot of value. We have a lot of innovation, but we can't just be making reckless bets. How do we fund this innovation without taking on unnecessary risk?

First, I would like to make a clarification, one that I think is important for iPaaS. Traditionally, these tools have been, sort of nice to have. They piece together systems, you can do it manually.

They automate systems, you can do it manually.

Now it has become a foundational technology, an enabling technology that will provide additional value with time, to a greater degree than it had before.

So what do you have when you have a foundational technology? You have rising opportunity cost with time. If you wait one quarter, the opportunity may decrease the value that you're able to see.

Forecasts are fuzzier than ever, but you have to act quicker in order to capture the, early lag of this value.

Integration itself is a multicycle asset, so you have to live with your choice of today's integration platform longer than you would for a lot of different apps.

Because of this, we want organizations to view their technology decisions specifically with foundational technology as continuous.

Remember that foundational layers rarely get swapped without pain, and a flexible iPass unlocks additional revenue streams with time.

The future value should likely be attributed a lower discount rate. What you're looking for in these solutions are ones where the magnitude of the value expands with time. Typically, with iPass and with innovation, this is hard. There are new paradigms that come up that decrease the value with time, and specific point solutions are typically designed for a specific integration pattern.

Adopting the solution that is specifically ready for AI is going to allow you to avoid this rip and replace and avoid the unnecessary costs that come with choosing an improper platform.

Now by contrast, when organizations do have to rip and replace, it usually signals poor decision making and poor processes in evaluating solutions.

Investing in dedicated process and collaborative decision making can significantly cut these costs and improve the value that these solutions deliver with time.

Now let's let's review.

Foundational technologies like iPaaS deliver immediate savings through their ability to cut costs and reduce the operational overload.

But their deeper value lies in on not in unlocking new revenue paths and agility with innovation.

As flexible modern iPass solutions make launching services faster and less risky through reusable connectors, governance, and monitoring, and now agent orchestration and management, innovation becomes simpler and organizations can adapt faster.

Because switching foundational layers is hard, selecting a vendor with clear AI roadmaps and open extensibility is critical.

Ultimately, your integration choice today determines how quickly you'll be able to adapt when market conditions or technologies eventually shift.

When evaluating candidates, you should ask, one, will this platform accommodate the next generation of AI and agentic models without major rewrites?

Two, does it support modular upgrades to avoid forklift migrations?

Three, can it scale economically? Does the value of this solution expand with time as event volumes spike, as integrations increase, and as the number of applications needed to integrate and agents needed to integrate increases?

Answering these questions upfront ensures the integration spine becomes a long lived asset rather than a short term patch.

In other words, your iPaaS platform will become a part of your strategic advantage, not just another tool to piece things together.

With that, I invite Talassie to share how some of these, examples come into fruition with real stories.

Thank you, Alex. And, that was a great, you know, research and, well put, just, you know, single word. We have to build for what's next.

Having said that, before I dive into the transformation journey, let me briefly introduce, you about, Yext.

Paul, next next slide, please. Yeah. Yext is the leading digital presence platform for multi location brands. It helps organizations manage and scale their, brand presence, across every digital touch point to be it in search, social, chart, listings, local pages, all from a single platform.

It brings, together AI powered capabilities, everything from generating SEO optimized, local content to handling reviews at scale. With generative AI, the platform ensures that companies show up consistently, accurately, and effectively everywhere their customers are searching. So, companies, you know, would not, you know, lose a customer, if they're searching for their brand. So that's that's the worst thing to, you know, that that should happen to the companies, and that is where we help companies, get connected with their customers. In short, externs your, you know, digital presence into a true differentiators, powered by AI, unified by one platform.

And, yeah. In my role, as a, you know, a lead for IT and business applications, so my focus, is on modernizing, enterprise systems, to drive agility, visibility, and scale, so that the business teams can, move faster, with, you know, automated workflows and, AI powered workflows. So that is where I'll walk you, through how we transformed our integration landscape, to unlock, automation at scale and accelerate innovation across the organization.

You know, with whatever, Alex mentioned in his research, you know, why we should look for a modernized, enterprise, system, which includes iPaaS. IPaaS is this key key platform, I feel, you know, connecting all these business applications, so that they can seamlessly, automate the workflows, for your business users.

So let's dive, you know, deep into what we have done, in that journey.

Let let's talk about challenges. Right? To know what, warranted as, I know, to move to Trey and, I know, pursue this transformation journey. So our, like many other organizations, our integrations evolved, reactually one team, one system at a time. We ended up with multiple tools across departments, with no centralized monitoring and heavy reliance on, custom built reporting, using BI and data warehouse, data warehouses and our integration engine engine, like, you know, the the previous one, you know. It technically worked, but the logging wasn't effective. We had to route logs, you know, externally, to understand, if there are any issues or troubleshooting, used to be, you know, cumbersome, when those, you know, broke.

And there was no, real time alerting, which is, which I feel is an important, you know, feature. So we built it, you know, ourselves, into into the platform.

And with IT, owning one part of the flow, analytics, another, and then there is operations, used to, own, you know, another piece of, the tool.

There was no end to end accountability.

What flows across, across platforms and there are there are manual handoffs especially for, you know, logs, error handling, meant we we were constantly, firefighting, I would say. So which, yeah, is a high, operational overhead.

Paul, can we go to the next slide?

Now the next challenge is innovation.

So, like, again, it resonates with what, Alex presented. I know, our previous, you know, platform was technically strong, but, I feel it was operationally rigid.

For minor updates, it required deeper, developer expertise, adding a new business system, which happens in the transformation, you know, journey for any organization.

The new tools are, you know, coming up. We need to keep, up to date where to to address, the business pinpoints across the, tech stack. So if you want to introduce a new business system meant, you know, long lead times, you have to customize the scripts, and you need to handle a lot of manual change requests.

So, through through reusable, connectors, I think, they existed, but modifying them, without breaking something else was too risky.

So most automation was locked behind, you know, the integration team, creating a backlog.

So empowering, business users was, basically out of question.

And, this wasn't a tooling issue. So it's more of a, you know, delivery model.

The platforms were enterprise grade, but they are not, agile. They were heavyweight, but not fast. So innovation basically slowed, not because lack of ideas, but because it's, you know, how hard it is, to implement, them, in that platform's, following slide.

Now growth challenges. Right, as we expanded, you know, we, made an acquisition last year, and, we will or we might, you know, do some acquisitions, you know, as as a company, eyeing growth, and, you know, compete with our, in in the in the space we are, working on. So the the the architecture struggle to keep up, you know, during this, you know, m and a or integration processes. So we couldn't scale elastically and tooling costs grew linearly with the data volume and, environment counts.

So increased, you know, pricing or cost, you know, if we need to expand our scale, was a big challenge. And there was no out of the box, centralized audit log or a health dashboard.

So when we onboarded, new more departments, the complexity grew, and, we didn't have, no, the visibility we need, to manage, the day to day operations.

And this keeps, you know, to keep these things running, we lean on work around scrapes, middleware patches.

But, that only compounded the problem. So we were using enterprise grade tooling, but maybe solving, the problems with a duct tape.

So now, next slide, Paul.

Question, Talase, you know, making a making a move right from an existing from your existing API management platform.

Right? May may some to it's it's no small task. Right? It and it's a you know, it's was there a you've shared some of the challenges around your growth and, you know, you had obviously a fragmentation visibility issues and just the need to move faster.

Was there kind of a a moment where it went from, like, thinking about it, to, like, okay. Now we need to make a move. What was that what was that exact kind of pressure point? Was was was this was it some trigger triggering moment?

Yeah. So, it's, it's just a continuous practice. I know as a IT leader, you you look at the platforms, you know, that are high performing and what is your, you know, goal, that you set for your, department for next two years, five years? And, what's happening, you know, in the industry, talk of, you know, AI or, and your business goals.

So are you able to support, these goals, with your current, tech stack and, how how fast you are, I know, turning around, these, you know, requirements that you are getting. So this, basically encouraged us to define the problem. Right? So what are your, you know, current pinpoints, with respect to cost, inefficiency, risk, and, you know, lack of agility.

And then we will, you know, we tried aligning that with the business goals. And, we now I know, decided, hey.

We should, go and look out for, you know, tool that can give us, you know, this business benefits that we are looking for, and then we started this, journey.

Excellent.

No. Yeah. So, again, speaking of the goals, right, like, maybe it's an extension to your, earlier question. So what do we, want from this platform?

So we needed, basically, we needed to pivot. I know based on I know what our pain points, we discussed the challenges, from being tool centric, to outcome driven, and, our North Star became clear. I know integration should enable growth. You know, it's it should not, limit it, which meant, you know, unifying the architecture across the business units, enabling self-service for business teams, hyper, automation, you know, for repeatable workflows and reducing IT friction to let ops, you know, move faster.

This aligns with, you know, what we are seeing across, modern enterprises, you know, we refer to a number of, you know, researchers like, what Alex presented, you know, a shift towards democratized automation, not just connecting systems, but empowering people to build, you know, with confidence.

A quick question, to ask you. I mean, I guess, all these aren't created equal. Right? You know, these goals, right, I mean, kinda going back to, you know, some of the areas where Alex shared as well.

I mean, there there is, obviously, there's efficiency aspects around, you know, getting your arms around again to more of a kind of a cohesive integration architecture, and then, obviously, you know, decentralizing and so that so your your business teams can build, you know, cost savings. Was there was there a was there a kind of a key goal, like an an overarching goal, like, you know, the the the big one? Right? That you're like, this is this is the must have goal for the for the for the project to to to make the move?

Yeah. So the the the bigger goal is, you know, hyper automation. Right? So where, you you you want to, automate, you know, the repeatable tasks again with efficiency, and how much, faster you are turning around, you know, the requirements and focus on strategic initiatives.

So freeing up, I know, some bandwidth, for for the developers so they can, they can focus on, you know, building what's next rather than, you know, trying to troubleshoot issues. So the bigger goal I would say is, you know, are we building, for future? So are we planning for future, the hyper automation, which includes, you know, this rolling out AI agents, rolling out automated, you know, workflows, for all the repeatable tasks. So to bring business efficiency.

So indirectly as a, business applications or an IT or what you are, impacting is, you know, the faster product, you know, rollouts, and, the customer facing teams would would would be able to perform their, you know, jobs with ease and without much, you know, running into issues, be it, you know, your sales teams, finance teams. So in the back end, our teams are empowering them to, to to to to make their, you know, job with much efficiency. So having a integration platform and doing what I mentioned, you know, around, in future, we are trying to roll out tidy agents, which I will, I know, speak about innovate and then hyper automating all these repeatable workflows.

It's it it was need of the hour, you know, if I have to mention.

It was really kinda like, I guess, speed and velocity was a key was a key goal?

Absolutely.

And, I mean, I'm curious. I I know you touched on kind of the kind of a one of the challenges being, being the backlog. Right? A backlog request. Were you were you seeing that backlog in the building, or was it, where where where was all demand kinda coming from on the, on the backlog? And then how did you see that kind of growing with with with AI and those areas of new business demand?

Yeah. So, see, like, like like we are planning for our teams, I think, the different business units have a a road map of items that they want to roll out, you know, be it changing the strategy or changing the, you know, selling process or changing the, you know, finance, you know, models, pricing, NPIs, new product rollouts. So there is a constant transformation happening on the business side, and we have to support them at the same pace. So these requirements keep coming, to us, you know, and they stay in our backlog. And we have to address, you know, based on prioritization, you know, there are screens and, stuff like that.

So, now if we if we, you know, fast forward, there are some requirements which, you know, the users can build by themselves. So, you know, a simple flow whereas in, like, alerting, when, a new deal, comes in or a deal for this. So these sort of small, workflows, the business users can, you know, develop themselves. You know, you have a connector to CRM.

You go log in and, keep an alert. There is a, you know, there is a, components that they do how to, you know, build a flow. So these things, the users can do it themselves. They don't have to wait, you know, for, for for the integration developers to come and help them.

So, that actually freed up some bandwidth, and we can focus on, you know, the larger initiatives and, help, build those automations or integrations, which are critical to the business systems.

That way, I think, there was this, efficiency, that we, brought into picture and, yeah. Overall, does that make sense? Let's Yeah.

Yeah. I mean, I understand you mentioned, and I I know Alex touched on as well is kind of future ready. Right? Is is the stack is the tenant you've got today, you know, the the right fit for the requirements you've got kinda coming down the pipe?

And obviously, you know, every every IT organization, every business is looking at AI. Right? Whether it's, you know, with, you know, a genetic or, you know, infusing anti business processes or, you know, you know, rag or those kind of areas. We're curious, like, you know, obviously, how did you see your iPaaS as an enabler, to to to help facilitate, some of the AI projects you saw coming down the pipe?

Yes. That's a very good, you know, point, and that's that's the happening, you know, discussion everywhere. I know, not in our organization, but whomever I talk to in my peer network.

So, this, so the agents are already here. I think comp some companies are, you know, adopting them, you know, much faster fashion. But, so the iPaaS platform is key, what I feel, to enable those, agents. So I'll I'll take a simple use case, and, I think, Trey, also implemented, you know, a ITSM agent, a IT ops agent, which can basically solve some of the, you know, issue, some of the issues that, your IT agents or, you know, the human agents resolve, you know, like access provisioning issues or deprovisioning or troubleshoot some of those errors.

So these things can be implemented now. You know, when you have, you know, right data and right API connections and you have a tool that can, support, you know, that integrations to different, platforms, you know, in a much seamless way. So these are, you know, becoming reality now. The same integration developers can, use those components and build a agent in conjunction with any of the, you know, AI, models that you are using.

So we can call, the the the API of the agents and then build a agentic framework to basically build an actual agent. So, which can think and, you know, troubleshoot and also action on a question. Right? So, there's a difference when we talk about agents, for the first, part, the chatbots which provide you, you know, right answer or right, data, which we are which we have, rolled out.

Again, apart from the IT agents, and this is becoming much much, common practice, within the business units. Earlier, you have to refer to, a bunch of I know a folder in a Google Drive. Now what we can do is you can connect that Google Drive to a chatbot, and you can ask a question.

The bot can just, you know, look at all the documentation and give you a specific answer, that you need.

So that you can extend it to to build an agent with some, you know, action where it can go to a connect to an application and perform a business logic and throw you a outcome, in a Slack channel or, in your email.

So that's, we we are already working on it, you know, for with with, a few of the models that Trey has, you know, built in into the platform. Yeah.

That's a that's a great point. I mean, I think some of the areas that the folks often don't think about when they're building agents. Right?

You know, beyond just knowledge, right, taking action, right, is really important. Right? So it has to be more than just simply search. You'll you'll, you know, the you know, in ITSM, for example, you take action.

Right? You know, creating a service ticket, right, or a part of the preset. Right? Actually acting on behalf of you and actually performing the task.

I think that there are other, like, endpoints, right, in terms of, as you mentioned, right, that, you might have an agent, but, ultimately, you gotta integrate the agent with a user Intelligence, right, whether it's the website or a Slack channel. Right? Your users have to better interact somewhere, like, conversationally or towards a triggered agents. And, obviously, data as well.

Right? The the agents only only as good as the data. Right? If you if you're gonna ingest a Yep.

Article source, right, or a narrow set of sources, your agents normally that, not not that equipment necessary for the right answers, as well.

You'll jump onto outcomes, Teleski? Obviously, I know you've been real busy over there. Yes?

Yes. I think we, spoke about a few of them, but, you know, just to quantify, the outcomes, you know, the transformation was tangible and we reduced our build cycles, from, like, say, three days, you know, on on an average to one day.

The business users could build their own, workflows now, you know, without filing tickets and, waiting for the developers to respond.

We, eliminated some redundant, you know, tooling and cutting some licensing and, you know, operation costs, so which brought in, sixty percent of, you know, cost reduction. And, we see it, you know, it's compounding, like, every, every day and every, you know, now and then when we, roll out some cool features, like, you know, say, these ID ops agents, so on and so forth.

And most importantly, the adoption, you know, it's I would say it's exploded, you know, the nontechnical users. They became, you know, a kind of automation builders for, you know, simple straightforward tasks. And, IT shifted, to architecture strategy strategy, rather than, you know, debugging workflows. So, basically, it was, not just a tool upgrade, but, like I said, you know, it was a operating model shift for us, which brought agility, you know, into the team, and then then the framework.

Curious on that that final area where, you know, you where you're growing your builders by maybe empowering, business teams to build. How important has that been for you to kinda reallocate your your developers to to other tasks? What what what's that meant for, for Yext? So rather rather than building integrations and enabling business team to build, and then, obviously, your your developers can then obviously focus on other things. Right?

Yes. That's, a very good, you know, shift or, observation, you know, which, we are experiencing now.

The freedom bandwidth, I wouldn't say, you know, based on, you know, the initiatives we are working on, we actually need, you know, more, bandwidth now, to to tackle those tasks. For example, these IT agents, I think, there's a lot of focus on their on on building those, you know, not just we just spoke about one example, for IT operations. So there are a lot of business operations where, you know, we can replicate, those agents and solve some, you know, problems with these agents, take care of some of the repetitive tasks, you know, with the agentic, framework.

So there is, I mean, we we need a lot of hands, you know, to build those, you know, robust frameworks and also, learn these models. They are evolving every day. There is something new that is coming up, you know, be it, Gemini, OpenAI, or, you know, Cloud. So we need to, keep keep up to date with those frameworks and, you know, leverage any new capability that they are rolling out.

So there is, a lot to do. Right? And then, for the business users, again, it's, while we are saving bandwidth, for the developers, the business users, again, they are finding more ways. You know, sometimes, they might see, you know, they might be doing those repetitive tasks and might not translate into the requirements now with the tool available.

So they can just experiment and build a simple workflow, for them so they can automate those repetitive tasks. So it's a, win win I mean, win on the both sides, you know, for us.

That's great. I mean, so so just just some just some outstanding results. I I'm wondering I'm wondering how how long it took you to build those hundred plus integrations originally, and it's pretty impressive to to to for, you know, hundred plus integrations migrated in, in three months.

Three months. Yeah. That was, even, we, in my experience, I think this is the first migration project that we, you know, delivered before the timeline. Right. I can say.

Yeah.

And, you know, extending there will be some I know always you see some, you know, issues. You know, we cannot plan a hundred percent, you know, perfect timeline for any of these migration efforts. But, you know, once the developers got, you know, feel of the components and the the workflows, they were, like, building those, super fast, you know, flows and migrating them, over to train.

Yeah. Well, I I think it's a great great segue, obviously, because I mean, let's say I mean, it's as you know, I think whenever, you know, you're in IT, right, and you're tasked with the migration. Right? I mean, there's a a small amount of sweat.

Right? You know, a little stress around that. Right? And, you know, make sure it goes smoothly and that, you've, you've you've you've you've you've met the requirements.

Right? And there's no, there's no mismatches between the tool you had and the tool you moved to. And I know you've you've learned you've learned some, pre learned some lessons along the way as well. I mean, I know that the best I think best practices will be, you know, ideal for our audience here.

Right? There's some things to to think about when when you make that move.

Definitely. Because, we were not just, you know, doing a a lift and shift, you know, from one platform to platform to another platform. We're we're utilizing that opportunity to, you know, learn and do things, differently and, you know, in a more better fashion Yeah. Than before. So what worked for us, you know, while we were, doing those, migration is, you know, for I think reusable components, I would, you know, give you top priority when you are, trying to, like, whether it is a fresh implementation or doing a migration.

Try to identify those, you know, early in the stage and develop some templates which you can reuse later. Or, we were talking about enabling the, you know, business users to automate their own workflow. So these templates can become handy when you are, you know, doing some training or, providing, you know, or creating an enablement program for them. So focus, I know, when you are, designing, your workflows, integrations, identify those reusable components and, that would, you know, come handy, in your journey. And the second one is, early engagement, with the business users. I think this is your opportunity to engage them again, to understand, the actual requirements.

There would be an opportunity to streamline some of them, optimize some of them, and improve some of the, integrations that, you know, or you might have developed, many years back. So, keep that, take this as an opportunity so it will help, for the future engagement as well. So in all business users, certainly, in the stage. And real time monitoring, so it's it's, you know, we are we have beautiful out of the boat, out of the box dashboards and, you know, the alerts, alerting mechanisms.

So visibility, wasn't a afterthought. It's it's basically a requirement. So when you build some integration, I think you should be able to monitor and, you know, identify any metrics or it should be easy to troubleshoot. And, the logging is critical for, you know, being a public company.

We want to keep, you know, the logs and then, you know, for any SOX compliance or audit, you have to retain them and use it, you know, for any evidence or, for any, audit and testing, you know, to some of the controls.

I'm curious on the on the the reusable components aspect. Obviously, a big focus here at Tre, right, composable Intelligence. Right? And we have building blocks so you can reuse.

Obviously, it helps with helps with the maintenance. Right? And also, I mean, ensuring that, you know, these templates are certified. Right?

You you know that, you know, that that strong that that that that proved out integrations that your business team can then take. Did you I mean, do do you consider yourself like a kind of forming of a center of excellence? Do you have a center of excellence kind of thing with, with with on the on the delivery side of things?

Yeah. So once, I know, like I said. Right? So we now we are working on a structured, a kind of enablement program where, you know, we create a role based onboarding and reasonable templates, some support and governance, like co billing workshops.

Basically, you need to create that process so that, I know the business users are not completely on their own. Like you mentioned, maybe we can call it as a center of excellence. So there is a, you know, team that is, you know, available, and there is some, you know, artifacts, documentation, these templates, and playbooks they can always use, you know, before actually reaching out, you know, for some co building workshops or, any other help. So definitely, these things would be, yeah, helpful once you create those structured enrollment program for the business users.

Got it. I know. I remember I mentioned six as, three of the best practice here, Jelasic.

Yep. So yeah. And furthermore, yeah. Continuously, optimizing the workflows, I think, especially when you are adapting a new tool, you keep, you know, learning, while you build, I think, over a period of time, it gets you I know, it becomes as a part of your, you know, rhythm. So every, the best practice I would say, again, not just during the process of migration, I think every quarter we, review and, you know, clean up any inefficiencies or, you know, any unused workflows so that, you know, the system is always clean.

And, self-service, like we spoke about, you know, creating those structured enablement program, we'll we'll actually bring in, like, you know, more power, to to to to the platform. And we we spoke about democratization democratization of the automation, platform. So that's the self-service piece and cross functional ownership.

Now, I know with, breaking this silos, the IT ops and, you know, data, they there's a lot of collaboration now.

Now they it's it's more of a shared goal, rather than hand off earlier. They all work together on on on bigger initiatives and, outcomes.

Next slide. I think yeah. We have ten minutes. So next steps, I think, having, made the, move, to Trey and, like, we spoke about few, during our, some of the while answering some of the questions.

I think, the the the major next is, you know, step would be, getting, in into a intelligent integration platform, right, which can support billing, IT agents for specific use cases, and then, expand, I know that trade, adoption across the departments, when we, finish this enablement program so we can roll it out to all the business users. And, yeah. Few of them that we listed. But what, I can summarize is, you know, we are laying the groundwork, you know, with clean architecture, observability, and modular design.

That's what, you know, will let us plug in into, you know, AI meaningfully, and it can become, you know, a multiplier for our, future road map.

Perfect. That's a that that that was a great, summary, Telesis. Thank thanks for sharing that, your your, your experience running through this at, at Yext.

It sound I think you've been pretty busy by by the looks of it.

We are. We are.

It's it's, I think, not just me. I think my peers are in our Yeah. Peer group, in different companies. I think with all the, AI and, you know, everybody's, you know, looking at future, billing for the, you know, what's next, in within the business organization or, you know, within IT. So how we can get those benefits, to to to, business users and then thereby, you know, providing that best, you know, customer experience, using our products. So it's, yeah, it's it's a great place to be in, with what's happening in the AI world versus, the business systems and combining, the best of, you know, both the worlds. It is a a great learning for, you know, everybody, in this in this area.

Actually, I got a I got a question actually. Actually, along some lines when you presented, questions that what what what are the queue as you get done done the requirements, what are the was that would that as you go through the evaluation, what was some of the key requirements, you you looked at?

Yeah.

I think, so we should define, you know, problem first. Right?

I think we were not looking for just, another tool. We needed a platform, that could scale, empower, and, you know, simplify some of the things. So, key requirements on a, you know, high level, you know, strategic, like, it should align with our, you know, business goals, like cost reduction, faster delivery or scalability.

And then, comparable, you know, options, based on, I think Alex presented, in his research around the total cost of ownership, right, and what's the ROI, and how strategically that will fit, into our tech landscape and, then quantifying the benefits. Right? So, the cost savings, time reduction, risk mitigation, and business impact. And, finally, then, you know, once you, we have those requirements, then the tool, you know, capabilities, around the reusable, you know, components that, you have, number of connectors.

The platform already has the built in connectors, because the business applications, it's it's very wide and, you have, like, if you bring in any quadrant, from any research form, like, you have twenty applications in each area and over a period of time, there are, you know, it, companies, you know, will bake in different SaaS platforms. So having those built in connectors will make your, you know, development, faster and those connectors should be efficient, so that you can automate any which workflow that you want to. And finally, what is, you know, we are building for what's next. Right? So be it, you know, agentic framework or hyper automation. So what kind of controls, the platform has, for developers if they want to experiment something, they can they build their custom connectors?

Yes or no? And what what is the feasibility of, you know, having, that kind of capabilities. And, also, I think, finally, the kind of partnership, you know, that is pretty much important as we continue this journey, you know, any which vendor. Right? So Trey is a great partner for us, you know, helping through the journey and giving us, support wherever we needed. So we needed that kind of, you know, customer support, you know, work, as a partner rather than a vendor.

So these were some of, you know, Yeah.

Yeah. And I think one one of the I guess I guess a follow on question. I think what we shared on the next steps, slide, I guess, about what what on the IT agent sort of things, what what I guess, question is what what you're looking to build on the I on on the IT agent sort of things?

What, I guess, looking for, like, what what what specific you know?

Yeah.

So, I think, we we we we touched upon, that little bit. Right? So, the, for example, the easiest example I can quote is the ID operations. Right?

So day day in day out, we see, a lot of requests, from the, employees and then we want to provide, you know, best employee experience by turning around those issues. So, very quickly. So now with these agents, you know, some of those, you know, simple issues, like like I said, like access provisioning or troubleshooting an error. So your identity, you know, management platform is not allowing to log in or your multifactor authentication doesn't work.

So these things are, like, most common category of, you know, issues and they have, a fix which is, you know, in our knowledge base or it's, you know, it's sequence of steps that the human agent performs. So we can automate that and bake into, the operations agent, so we can reduce, you know, some of the load on the agents so they can focus on, more, you know, infrastructure related, issues supporting our product teams, engineering teams.

So that way, I know that is one use case. And second use case is, you know, there are a lot of repeatable work workflows. Our business users are, you know, using, you know, still, I think, spreadsheet is a big, you know, contender for any automation you try to do. There are a lot of, data analysis happens in the spreadsheets. So there are a lot of formulas sitting there. So, can we build, you know, Intelligence that can, you know, help business users to interact with that, you know, data and, provide some suggestions, provide some trend analysis, provide some, you know, maybe, making some kind of forecasting. So those kind of, you know, kind of r and d we are doing, where where we can, get some requirements.

And, third one, again, and we have three minutes, is something we are learning from the noninteractive, you know, use cases that, the business folks are using. When I say noninteractive use cases, like, what questions they are asking, you know, in in the stand alone, you know, whether it is ChargeGPT, Gemini, or, you know, AI platforms. What kind of questions they are asking and what kind of, you know, problems they are solving there. So that, becomes, you know, a requirement for you when you are building, IT operations agent or any other AI agents. So whatever the stand alone questions they are asking, so we can try to take them and build agent around it, which can basically sit in the back end and process these, requirements and provide, some actionable outcome. That makes sense.

That that that's great. That that's great detail.

Deliberately, you broke it down. The, as you could Intelligence more one more question. Alex, I I guess time to get your crystal ball out.

What, guess, where where do you see iPaaS heading on next year?

Yeah. Absolutely. Like I had, mentioned at the beginning here, like, every space we're covering here, there's a lot of change going on, both internally with the way that these platforms are using AI in order to integrate into their own user experience, what are developers doing within these platforms with AI, as well as how AI is impacting integration itself as a landscape.

The recommendation that I would give to everyone who's currently working within iPaaS, is to really own your process and understand the tools that you're using, the user surfaces you're interacting with, and how you can piece all of those together at the best.

So we're gonna look for a lot of change. We're going to look for a new class of usability that's really leveling up the natural language interfaces that we're seeing, and we're going to see a new class of functionality around managing AI agents and these processes with AI. So it's gonna be a lot.

Very much look to your vendor road maps, look to vendors who are, interacting with you and taking a customer first focus in order to adjust with this innovation.

Wow. Perfect. Thank you so much, Alex. And, also thank you thank you, to Lassie, as well.

That was our, great event with a lot of practical tips. I wanna just share some, just share some next steps. And, you know, we have a we have a mix of folks, on the, attending. Everyone from, folks that knew the trade and also folks that are, running trade, today as well.

So if you're if you're, if you're running trade today, you welcome to reach out to your account manager, and we can add add, agent builder, to your, to your tray, to your tray deployment, your tray workspace.

If you're not a customer, and you wanna get learn a little bit more about tray, just drop us a line, in the chat, or you can head over to trader a I slash demo to request a a personal one one demo.

We'd we'd love to learn about, your projects and, what you're doing. Everyone who's attended, will get a complimentary copy of our AI agent playbook.

Provides basically real kind of practical tips on the strategies on how to how to get a successful agent deployment, and and kinda break the ice for your first deployment out there.

And, also, you can head over. If you'd like to get a copy of the, the matrix that, Alex shared, on his presentation.

You can get the full report as well. It covers all events in space.

And there's the, the link placed on the bottom right hand corner there that if you wanna head over to get that matrix from, the tray, website.

So I wanna just, thanks again to, Alex and to Lassie. Really appreciate sharing your insights.

Best practices, you know, where the industry is going, well, you know, all all the good things, drilling down into agents and getting specific about it as well.

It's a great session, and, also, I appreciate everyone attending as well. And, so have a great afternoon or or evening.

And, thanks, and goodbye.

Thank you.

Thank you.

Featuring

Tulasi Donthireddy
presenter

Tulasi Donthireddy

Senior Director, IT & Business Systems

Yext
Alexander Wurm
presenter

Alexander Wurm

Senior Analyst

Nucleus Research
Paul Turner
host

Paul Turner

Automation Expert

Tray.ai

Let's explore what's possible, together.

Contact us

More like this