Software systems sit at the heart of every modern enterprise.

Far too often, your ability to break down the silos between sources of truth and action are limited.

Moving to a world where anyone can solve problems without the constraints of technology, business moves faster, customers are happier, employees are more productive.

Today, we're delighted to introduce Tray Merlin AI, the first ever natural language capability for iPaaS that anyone can use.

Merlin is seamlessly integrated throughout the Tray platform, aiding employees of all types to obtain answers at the point of decision.

For technologists, it simplifies the process of creating automations, allowing you to build faster and better. With Merlin, you can keep pace with a rapidly changing world, unlocking the potential of the automated organization.

We are excited to show you what Merlin can do. Please welcome our host for today's session, Tray's CEO and co-founder, Rich Waldron.

Good morning. Good afternoon. Delighted to be here today to introduce Merlin to the world. Let's kick things off.

So to get us going, I actually wanna take us back to the start. You know, our founding perspective when we started building Tray still rings true today. The things that we're hearing when we're speaking to customers or prospects and we're out in the market, you know, I'm looking for a low code platform. I don't code, but I'm technical. We need one place to build all of our automations.

These are the things that we set out to solve when we started building Tray.

The big thing for us was how do we take the power of automation and put it in the hands of a much broader audience? How can you take the flexibility, the governance, the scale, all of the things required to successfully build a powerful automation and make it accessible to the rest of the business? How could you take this visual interface and enable people to do things that they historically could not?

And so when we think about the iPaaS space itself and we think about some of the challenges that are being solved by iPaaS today, you know, the market has changed pretty fundamentally in the last couple of years alone. You know, what we're seeing today and the reasons that people are coming to Tray to solve these problems are that they're stuck, you know, with rigid services that legacy iPaaS isn't allowing them to unlock. They've purchased lots of software. They've gone through digital transformation processes.

They now have these microservices architectures. They've got data that exists in silos all over the organization, but it's so hard for them to be able to do anything with it because the tools of yesteryear weren't built to solve the problems that exist from the cloud today.

And then just to twist things up entirely, we've now seen this huge, evolution with AI that's completely changing the game right out in front of us. And so the combination of these things means that prospects and and and business owners and, you know, folks that are out in in the market today are feeling this pain on the front lines. They're feeling the pains of not being able to solve the business problems that are right there in front of them because they don't have the tooling to be able to get things going.

The solution to this problem, as we see it, is composability.

This thought process that you can form larger business capabilities by combining smaller building blocks. Now if you think about the way in which, your own business operates or the way in which the market changes in with its ferocious pace, you need something that is always ready to scale, that is always ready for change, that can handle, you know, the fact that everything occurs now in real time. And to do that, you need to be able to build solutions that you can, morph and change and adapt dependent on the business challenges that you're facing in your day to day life.

Our take on this is something that we call the composable future.

And for us at Tray, you know, we see that there are three kind of core concepts behind this future. Firstly, thinking about building business capabilities using the full software stack. That means having a platform that can power and push data between all the services that exist in the market today.

That means in the middle layer, you've got something that allows you to, you know, produce that logic to connect the different services, to be able to do the things that you wanna go and do. And then lastly, it's all about that interface. How do you interact with it? How do you make it available to others?

In the same vein, nothing should be built from scratch. You know, everything should be reusable. All of the work and everything that we put into the years that we spend, solving problems in our day to day job, can be reused. They can be rebuilt.

They can be repurposed. Everything that we're learning along the way or the experience that we have gives us this capability to go and solve the problems in a completely new and novel way. And the last piece of the puzzle is that you can do it in the way that is most natural to you. That to me is critical.

If you think about, you know, the different roles that we have within organizations, you've got a combination of different skill sets from the very technical to the not so very technical through to those that, you know, really understand the business value of the problem that they're solving or the challenge that they're facing. And when you can kinda mix and match the skill set with those that understand the problem, you end up with amazing outcomes. And so finding solutions that enable you to solve these problems in the most natural way is key to this concept of the composable future.

The great news is we're already seeing these results. You know? The problems that we've highlighted here in the market that is starting to be solved by iPaaS today, you know, we're seeing from within our own customer base real ROI, that can be measured in business outcome. You know? The speed of marketing velocity at Jellyvision, the cash flow savings that Mixpanel have been able to action, the speed at which Auctane can process their employees and provision them and get them going on the frontline.

And then lastly, Vox Media, being able to set up customers so much faster. All of these things have been enabled by having a composable mindset, by being able to take existing components, by being able to connect different data silos internally, by being able to, you know, bring a different experience dependent on the skill set of the team that is, working on these problems, and actually, you know, deploying these things and getting them into the frontline in rapid time. And when you go and do that, you know, the outcomes are amazing. It's not about, you know, are we gonna go through an expensive implementation process and figure out if this technology has worked in twelve months' time? No. We're getting instant results. We can quickly determine whether or not the things that we're bringing to the market today are making a difference.

But for us and for the rest of this market, you know, the game has just totally changed. AI is really the new automation opportunity.

The things that have held back the progress of iPaaS over the last few years, are blown away with the evolution that we've seen in AI. You know, now the speed at which you can deliver your integrations has gone through the roof. Right? This concept of, you know, being able to use AI to help you build faster, to be able to provide insight, to be able to support and figure out what's going on with your integrations speeds up the velocity in which you can deliver a solution.

Secondarily, it expands your pool of builders by being that kind of helper that sits on the side, that that person that you can interact with that allows you to build more comprehensive, more intelligent, and more efficient workflows means that for the first time, you can expand the pool of builders that exist within your organization.

And then lastly and critically, for this to really permeate across, an organization, you can now deliver automation to the frontline. This kind of natural language process that we're starting to see exist within automation means that you can instruct the service to act in a way that you couldn't before. It doesn't require you to go and build out. It means that you can get started in a way that is most natural to you, which aligns perfectly with what we just highlighted in the composable future.

So what we're gonna walk through today with Merlin is something that we call conversational automation. This is a new type of automation that is infused with AI to power three key experiences.

The first one is the technologists. You know, these are the folks that are building and iterating processes, using low code today.

With the introduction of AI, they can build faster. They can build more effectively. They can build more efficiently.

This really opens the gambit for them to be able to create and deploy faster than they ever have before.

Number two, managers and frontline workers can actually shift from doing manual work to getting automated answers and getting actions on demand. As you'll see very shortly, you can actually have a conversation with Merlin. You can tell it to go and carry out some tasks for you, and off it will go and complete that work. It doesn't require you to figure out, you know, how you're gonna construct the workflow or build out, using the low-code builder.

It doesn't require you to be an engineer that needs to go and write code and deploy it somewhere. For the first time ever, you can actually have a conversation with this. And then lastly, for developers, being able to ingrain their apps with AI-powered integrations and automations and build experiences around these, speed up the way in which they can provide integrations and automations to their end customers. So you can see across the entire spectrum, this conversational automation is really gonna change the game for iPaaS.

So the applications for a tool like Merlin are practical and endless. You know, you can see here, there are many different roles that are benefiting from being able to use the solutions that we're able to walk through today, from IT ops to the CMO through to a sales AE, through to a sales, to a service manager.

The different types of interaction, the different types of experience, and being able to use the skill set that you have and create a solution in the way that's most natural to you opens up a whole new opportunity for how you can introduce automation within your organization.

And so I think we're all ready. Right? We've we've heard enough from me. We've we've we've got had an introduction and a perspective on the space and how AI is gonna make things happen.

But what we're really here for is to get a demo and to see and meet Merlin for the first time.

So I'd love to hand over to my co-founder and our CTO here at Tray, Alistair Russell, who's gonna unleash Merlin on the world.

Ali, over to you.

Thanks, Rich. Yeah. I'm really, really happy to be here today as well, to bring Merlin to the world, to really show what we've been working on for a while now here at Tray. Something that I personally have been very, very passionate about.

I'm really, really excited about this as well. It's it's kind of building on top of the the great, sort of, you know, work that's been done in the last eighteen to twenty four months in the sort of AI and ML area, like, especially the sort of, you know, the progression of, of large language models and the capabilities that they offer. And really bringing the sort of the power of those and and and bring that together with the power of the Tray platform, such as our kind of, you know, hundreds of, sort of standardized connectors or our standardized APIs and that the sort of scalability aspect of the Tray platform.

Really bringing that together in, as you mentioned earlier, what we call conversational automation. So, it's, it's something that I, you know, think is gonna be, sort of a complete paradigm shift, both our industry and many other industries as well. And it's also only the tip of the iceberg. We're we're talking about, you know, sort of what's available today is already incredible, but it's only the beginning.

We're gonna see Merlin and other sort of, you know, sort of large language models and features built out over the next, you know, couple of years that are really gonna change the world and change how we interact with technology, how software is built, and things like that. So, let's get started. So I'm gonna demo a couple of the interface today. First one is Tray Chat.

So this is our kind of, conversational automation interface for for business users.

I'm obviously, you know, sort of high, high builder score if I you know, I don't wanna boast about it too much. Technical user, you can build, you can Tray, you can write code as well. But I still find, there's lots of value in being able to use a a conversational interface like this because it allows me to kind of, you know, ask questions and have a more natural language conversation and and get kind of answers and get results and actually action things, sort of, you know, on using that interface as well. So let's get started. It's built on top of, large large language models, like such as OpenAI, and you can ask it sort of simple questions like, you know, who are you? You know, it's it's self aware. It understands, you know, who it is and what it can do, and you can have sort of a natural conversation with it, ask it like, you know, simple questions like this.

It's gonna respond with, hopefully, yeah, I'm a Merlin, a helpful AI assistant.

You know? But you can also tap into the sort of the more kind of powerful aspects of large language models such as, you know, translation. So, you know, what is, hello, Merlin in French. And that will, obviously respond back with, you know, my French is rusty, but I'm pretty sure I understand that's what I'm I can validate that's, that's correct.

And, you know, but the real power with Merlin and with conversational automation comes when you start bringing in the SaaS tools and the ability to kind of interact with with other services, other APIs. And that's where Merlin really sort of, you know, really adds the sort of the value here. So let's take a simple kind of, sort of, you know, request such as, you know, look up, my email address on it, Tray on here that, which is a business intelligence tool. So what's happening here is Merlin is sending this request to the large language model, breaking it down into a number of tasks.

And then Merlin is the is the sort of, you know, the go between that basically figures out what connectors on the TriNet platform are useful. We'll do things like, you know, sort of ask for authentications. So you can see here it's requesting me, to provide a clear bit authentication.

I've already got one. I can confirm that, or I could create a new one if I wanted to. I can do that as well. And then it's gonna ask me to confirm that I want to do this request. So from a security perspective, we wanna make sure that we're not doing anything completely automatically. You know, you always have to confirm that these are the steps that you want to take. Now there's only one at the moment, which is, you know, mentioning that person by email.

I can click confirm. And what Merlin is now doing is it's now using the live language model to generate the connector calls. And it's actually using our connectivity API, which is an API that, we have customers using for other other sort of purposes. But one of the best you know, the great things about APIs and and, you know, sort of, the Tray platform is that it's flexible enough to kinda do anything.

So we we've repurposed our connectivity API for this particular use case. And so it's made a call to that connector. You can see there's lots of data. It's got the, you know, my name, location, employment, Facebook details, GitHub details, actually.

And that's returned that data. We've some Merlin summarized it, give me sort of all that relevant data. I can actually copy some of it as well. But let's, let's look at another use case that's taking me a bit more interesting. Salesforce, for example. Let's say, you know, get all open ops on Salesforce.

So, again, one of the great things about large language models is that, you know, they can take Sang and different ways of writing things and reinterpret them in the right way. So in this case, ops isn't a concept that exists on Salesforce. You know? It's opportunities. But the large language model in Merlin realized that, and has translated that into get all opportunities, open opportunities in Salesforce.

So let me confirm that and, again, confirm the sort of single, the single single task is gonna go and execute. And, again, it's going away, making a call to Salesforce to find the records. It's gonna come back in a second and, summarize the data that it's that it's found.

At the moment, what it's doing is basically configuring the connector call. So it's taking our standardized schema, so the input schema for the Salesforce connector. It's already it's already identified the connector to use and and the operation to use. That's what Merlin's sort of done already.

It's now basically, you know, configuring that connect call, making sure that the API inputs are sort of, you know, correct, and then they're making the call as well. And then that call will return some data. So you see there it goes. Took a little bit longer that time than previous quarter to clip it.

And it's returned thirteen records. I can see all the data stack and copy the the entire kind of payload. If I wanted to choose a clipboard, paste it somewhere else, I can, you know, scroll through. I can copy, you know, some of the values like, you know, individual values like names, etcetera, navigate through the pages.

But the real power comes when you start to sort of, you know, to hook multiple steps up. So it's, you know, it's great that you can go and get this data. You know, fantastic. But let me do something with it.

So, let me summarize it. Summarization is one of those things that Merlin, you know, allows out of the box really, really easily. So let's, summarize the ops by stage and then add to a new Google Google. Can't spell Google.

Sheet called queue to open ops.

Right.

So, again, you know, I've used the English summarize with an S, but Merlin is gonna actually translate that into summarize with a Z.

But it's you know, the one thing it's doing here is it's I could have done the entire kind of, request in a single kind of statement, so I could have asked it to go to Salesforce and get stages then summarize, etcetera. But what it's actually doing here is actually using the previous data. So we've already got the data from Salesforce, so we don't need to go and make that call again. So here, you can see it's asking for my Google Sheets auth.

And, again, the two steps that it's gonna do are two new steps. It's gonna use the data previously to summarize, and then it's these two new steps it's, you know, sort of confirming that I want to do. So create a a new spreadsheet on, Google Sheets and then add values to that spreadsheet. So, what it's doing now is it's doing a summarization step.

So, again, one of the really powerful things in Merlin is it can do things like summarization. It can do kind of maths and logic and things like that. These are things that the large language models can't really do, very easily. Some of the sort of especially the sort of the math doing you can do sort of simple mass certainly, but not more advanced mass.

It doesn't understand things like current date and stuff like that. The data it's rooted and trained on is in the past. So it's now created a new Google Sheet, Q2 open ops, which is great.

And then let's, see, it's adding the values to the spreadsheet. So, again, it's, you know, what it's doing now is it's going away. It's, you know, taking the sort of the input, which will be the, the sort of, you know, the summarization and the the the sort of the previous step. So it's it's using the ID for the created Google Sheet, and it's generating a new payload to basically call Google Sheets.

And there it is. So, it's finished processing the process in the request. It's got a link here. If I click open, there should be, there you go.

That's my new Q2 webinar. It's gonna go.

And then, basically, it's called the data in. But I maybe wanna do something different with that. So let's let's say, okay.

Let's send that link, to the Google Sheet to team at Tray, using Gmail.

So what I'm doing now is I'm sending this to, to my leadership team, you know, because I want them to see the open ops, and, I could have also actually included, maybe an agenda for our leadership meeting in the the body of the email and stuff.

But this is now gonna be, again, translate that into a couple of requests. So it's basically gonna send a link. So it's gonna use, like, Gmail connector. Again, I've already got an authentication there to confirm.

And then that's what I wanted to do.

And that's gonna take that link and it's gonna go and actually send it to Gmail. I could have also done it to Slack, for example. I've got a Slack channel open just up there, which I could have done that too but let's leave it just to Gmail. There you go. So I sent that email. It's the ID of the email. Results sent, and the exec team will now be receiving a link.

I'm in the exec team, so I don't think I can actually see that email, but I'm pretty sure that that's definitely there.

One last thing I'll you know, I always love to sort of play around, so let's let's do this. So, you know, write a poem to Merlin AI in the style of Doctor Dre.

It's always, it's always fun to to sort of, you know, just get it to do random things, like you would. I always ask my colleagues, work via Slack and stuff to write poems, in the style of Doctor Dre.

And I've done this a few times, and Merlin's produced here's some quite different results, quite interesting lyrics, certainly. Always good fun there.

While that's generating that, I'll start to think about moving on to what we call Tray Build. So Tray Build is our workflow builder.

And where Merlin exists, what we're doing in Merlin there is we're basically using Merlin to augment the build process, basically, speeding up the sort of, you know, the ability to kinda build workflows. There you go. So, you know, a nice, you know, I would send it to people or so maybe I'll send it to Slack or I'll save that later.

So, let's move on to the builder so I can go and open the workflow builder here.

So this is, you know, Tray Build, or the demo of sort of, you know, Merlin that exists within the Tray builder itself.

Now, obviously the target of this is for people who, you know, are capable of building, who wants sort of, you know, to be building the workflows themselves and actually kind of have a persistent kind of asset that exists after that kind of initial conversation.

Now there's two kinds of scenarios where it's really incredibly useful. One is, for new users, as an educational tool. You know, you can basically ask it questions. You can get it to do stuff and, you know, get it to build workflows for you.

You can see how it does it. You know, if you don't know how logic works within the Tray builder, how some of the connectors work, you know, things like that, then you can get it to do stuff for you. You can describe what you want using natural language, and then you can see how that translates into an actual workflow. And then there's obviously the, you know, the more power user as well.

So the power users themselves, you know, like myself, you know, like, still benefit from being able to do things like this. So, like, you know, the idea is that Merlin sits alongside the ad step dialogue.

You can basically, you know, come in and ask it a question. This one's a bit more of a complex kind of description of what it wants from workflow.

I'll let that sort of get going. And, again, it breaks this down into a set of tasks, which in this case, the big difference between, Tray Build and Tray Chat is that it generates actual workflow steps. It doesn't make the calls to the APIs. It generates workflow steps that you can then run on a schedule.

You can, you know, run whenever you want. You can inspect. You can add. So it's about sort of augmenting your build process.

And, so you can see it's breaking these down into a number of steps. So creating a a Google Sheet, getting the leads from Salesforce.

So in this particular case, what it's doing is, it's basically getting all leads from Salesforce with a sort of resource web.

It's then adding them to a particular Google Sheet. So it's gonna ask my authentications now. We do this for two reasons. So, one is that, you know, obviously, you need authentications to run the workflow, so it'll set them at the time.

And if there were existing steps within the workflow wouldn't actually ask for those authentications again.

But the other reason as well is because quite a lot of our connectors, you know, use or sort of allow for dynamic calls. So, like, there's dynamic data lookups, which are the drop downs you see in the properties panel. There's also dynamic output schemes as well, and that data is useful for Merlin to understand how to do mapping better. So, you know, Salesforce has custom objects, custom fields, things like that.

You know, the static schemers that exist around state you know, Salesforce APIs, won't really sort of allow you to do more sort of powerful, sort of, you know, use cases and sort of powerful configuration of workflows. Now, obviously, this is, you know, a more complex statement, so it's, you know, taking a couple minutes to sort of, you know, to achieve. But in the background, it's actually already doing this. We're we're hoping to add, some more kind of iterative, sort of experience where it'll add the steps, one by one as it sort of starts to do that rather than, you know, wait until the end to sort of add all the steps, which it's just done.

But even still, you know, this is this is added four steps. It's done the data mapping. It's done, you know, done the configuration. So you can see here iit's named them all correctly as well.

It's giving them good, you know, useful names.

So in this particular case, it's, you know, set the title of this spreadsheet web leads. It's added a sheet with the title today. It's done this Salesforce lookup. It's, you know, set lead.

It's added condition, you know, lead source equals, web. It's done the mapping between, the loop step and the the Salesforce step in my records field. And, again, it's coming to here. It's like, you know, using the spreadsheet ID and the worksheet name from the first step.

It's basically looping, you know, using the sort of, you know, the backing from the, the sort of the Salesforce step as well. And it's done all that automatically. Now this in itself would have taken me, you know, would take me three or four times as long to do this. Like, you know, it might seem like a simple workflow, but, you know, it's still gonna take me ten minutes plus to sort of, you know, to get these, like, all configured.

You know, and that's if I know the the connectors very, very well, which I obviously do because I, you know, I use Tray all the time and especially these connectors, the ones that I demo with quite a lot, so I know them very, very well. This still would take me a lot longer, so there's a lot of value in doing this. But, really, what we want to do is we want to start adding, you know, the ability to kind of do more significant changes within this sort of workflow. So we've got a lot of, features coming up on the road map and the short term road map as well.

You know, the ability to kind of, you know, you can start adding steps in here. So you click, for example, say, you know, sort of, send, send a message to Slack channel, and then demos. And that would add a step within that loop. So you can then kind of add, you know, iteratively add steps in the workflow.

One of the things we really wanna do is the ability to kinda just edit steps that exist. Do things like data mapping. You can sort of say, okay. Map the data between these two steps, and we'll just take the two steps that already exist, and then just, you know, add them into the workflow or add the data mapping between them.

Things like looking up logs, we've been sort of playing around with these features, in development as well and the ability to basically take logs, and analyze them, you know, the errors and stuff and then suggest, you know, sort of updates and stuff to the workflow that we could do as well. And all of this really is an incredibly sort of powerful sort of foundation that we can build on because, you know, building workflows in this way is gonna just sort of make everyone's life so much easier, and the sort of features and the capabilities just gonna keep getting better. As I said, like, you know, this is the this is really the tip of the iceberg, to be honest. But now if I, let me move back to the deck. So I've got to get through a couple of slides before we move on and introduce our customer guest on the panel.

So let me just switch.

Cool. So just moving on, I just wanted to, you know, sort of cover some of the things I've already talked about today. But, you know, we very much see Merlin as this kind of, sort of layer that exists across the entire Tray platform. You know?

So we've got these three kind of interfaces that we've been talking about now. So our, you know, our APIs interface, our workflow low-code builder, and our Tray Chat, which is the new interface that we just, showed you today. Workflow builder and Tray Chat are powered by Merlin, that Merlin sort of weaved in. But, really, Merlin is gonna sit across the platform and allow you to do lots of things and how to bring that kind of intelligence layer into everything that you do within the Tray platform, whether that's, you know, finding templates, whether that's help documentation, whether that's analyzing your logs and analyzing your sort of, you know, data usage and things like that.

There's lots of things that sort of we have planned and stay tuned for the upcoming road map on Merlin.

But it's all built on top of, obviously, our enterprise core from a governance perspective, security perspective, all those sort of, you know, things there and all the capabilities of the Tray platform, you know, the connectors, the sort of workflows, the sort of, you know, the config wizard, etcetera.

And it's only really possible because of the way that we built the Tray platform. So, obviously, you know, API first, we have APIs for everything. I mentioned earlier that we use our connectivity API extensively within Merlin, and that's like a standardized interface to all of our connectors. You know, we have the modern workflow builder itself, which, you know, is very easy to integrate directly, like, you know, integrating that AI experience within the workflow builder itself, which, you know, is fantastic.

And the big thing is the connector library. You know, we have hundreds and hundreds of, you know, standardized connectors with standardized interfaces that make it really easy for OpenAI and the large language model to help us generate the right configuration, the right input, and really sort of understand which connector is the right ones to use. And, of course, the scalable serverless architecture. You know, this feature brings a whole new type of user to the platform, And our scalable service architecture is really gonna allow us to scale and support this sort of, you know, the growth that we're gonna see over the next few years because of features like Merlin.

I mentioned again already that we very much think of Merlin, as the the body and the manifestation of this conversational automation. You have OpenAI, which is conversational intelligence. You know, the ability to have a conversation that comes across as intelligent. It seems like you speak to an actual person.

Conversational automation is like having a digital teammate. It's like being able to just have a conversation and say, can you go and do this? And having someone that can just go and do it. And one of the real benefits is the fact that that person doesn't get tired, you know, understands way more than most, you know, sort of teammates that you probably have about all the different services.

And you can also, you know, have almost an unlimited amount of them. You can just go far off, like, you know, a hundred, you know, parallel processes to go and do different things. It's like an army of people who can just go and do all these small tasks. So the future is, you know, the future is gonna be very, very, you know, sort of interesting as we evolve these sort of, you know, these tools and these features to really help us to build software and to sort of, you know, automate our lives and automate our staff as well.

Quick view of how it works. Again, I've already kind of, you know, talked about this a little bit. You know, Merlin very much is this kind of middleman, between OpenAI and the user, you know, sending the request to OpenAI to pass what needs to be passed, you know, sending them back and, you know, identifying connectors, identifying or, like, you know, making the calls, etcetera, summarizing the results, etcetera. So it does a lot of work in the middle and uses OpenAI.

And one of the things that is, very important is the governance aspect. So like I said, we don't send any data to the large language model. We've been very intentional about this that we don't, you know, send the user data there. And we've done it in a, we built it in such a way that it only needs to send, sort of standardized kind of schemas and things like that to generate the right connect calls.

It doesn't actually need to send the data from the customer over to the the large language model itself. And, of course, it's all built on, as I mentioned, our enterprise core, you know, the same sort of governance permissions, audit trail, all those sort of things that you can expect from the Tray platform already is built into Merlin. So, you know, users can't use Merlin to do anything they couldn't do within the builder. You know, we provide full audit trails, like, you know, logging, all those sorts of things.

So you can understand what people are doing, give them sort of certain permissions and stuff like, you know, use the workspaces feature to make sure they didn't have access to sort of, you know, certain authentications and give them access to do anything.

It's all, like, incredibly secure, scalable, and, you know, like I said, instrumented.

So we we really care about the security of this and sort of the governance, especially, from an enterprise perspective.

And that’s it from me, that's the demo. I hope you really enjoyed it. I'm really excited for the future, and how it's gonna evolve, and I'm really excited to get more people into the product using it. But speaking of someone who's been using the product, I'll hand it back to Rich who's gonna, certainly, you know, introduce, Steven and, one of our customers who's been using the sort of early version of Merlin for a while now. Thanks. Over to you, Rich.

Wow. Thank you for that, Ali. That was amazing. It was so exciting to see all of the different ways you can interact with Merlin and how it's gonna speed up the way that we can deliver automation across so many different departments.

I'm delighted to be able to welcome a valued customer of ours today, Stephen Stouffer, who's the VP of digital transformation and innovation at SaaScend. Steven is gonna be fantastic for us to have a quick perspective on AI as he's been someone that's been playing around with this for some time now. Stephen, I'd love to welcome you to the to the conversation. Come on in.

Thanks, Rich. It's good to be here. Thanks for having me.

Really appreciate it.

So, Stephen, you know, as somebody that has been a Tray customer for many years, somebody that is a super user on our community, somebody that has built so many different automations in your time. I'd love to hear how you've been using AI in your day to day in in in your role at SaaScend.

Sure. Yeah. So it was funny. I recently built something just kind of out of necessity.

Many people don't know this about me, but I'm very dyslexic.

And I was scrolling through the OpenAI documentation as you do when something like this comes out. And I noticed they have an edit API. And that got me thinking. I'm like, could I incorporate this edit API with Slack?

Because I sent so many misspellings via Slack. So I put on my Tray builder hat. I looked at the API documentation and I built a little Slack app that basically sends every single Slack message, whether or not it's a DM or in a channel. I funnel it through the edit API through open API powered by Tray, a web trigger workflow, and it actually corrects my spelling on the Slack side.

So that is a very basic use case there. And then I and it worked really well. And then I thought a little bit more about that. I'm like, you know what?

If AI can make sense of what I'm saying, I'm wondering if it can also make sense of what my customers are saying. So we have a lot of social media integrations for lead generation.

For those who integrate with LinkedIn or Facebook, you know that there is open text fields for state and country values. So a lot of folks put in state codes or full state names or country codes or full full country name, or they can just misspell it. Add an extra s to the end of Texas. And all of that breaks lead routing.

So, I, went back to Tray. I built another workflow similar to my initial one with Slack, and I actually started funneling all these data values in through AI and it's been just correcting them for me. So, in the old world, you'd have to go in there and manually adjust these misspellings. But now powered by AI, it all just fixes it, and then the lead gets routed directly to sales.

So, like, those are two very simple use cases, but, like, so powerful when it comes to thinking about, you know, time to sales.

Yeah. That's amazing. And I think the thing that strikes me about that is how that's kind of now a foundation for you to evolve from. So as you start making and and impacting your lead cycle like that or you, you know, start kind of correcting your spelling in the way that, is gonna impress everybody else within the organization.

The evolution from there is the exciting part. Right? Because the, you know, the business will change over time, and the data that you wanna capture from a lead will change over time or even, you know, the forms that you're putting out or even the destination source that that's going to. And so at every point, you've got something that you can kinda iterate from and iterate off to. And when you infuse AI in that, you know, it kinda blows my mind that the speed at which this stuff's gonna develop.

When you think about that, where do you think the biggest sort of potential is for iPaaS? You're someone that's been, you know, using tools like this for some time. You've already started, ahead of even, you know, the Merlin release, which I know you've been you've been using as a customer. Yeah.

You began, you know, playing with AI yourself. So where do you see the kind of potential for iPaaS, and what are you looking forward to doing more of?

Sure.

So technology is moving really quickly.

It's getting increasingly more difficult for businesses to basically just to keep up with that movement of technology. And I think when it comes to iPaaS solutions, and AI, it's a very logical kind of next progression with iPaaS to kind of incorporate AI into it for things like data quality, proactive monitoring and alerting, security and compliance, and then self healing remediation.

Those are all places that I think have the most potential. Probably the last one is the one I'm most excited about is just the, you know, as a human, I'm working eight hours a day, maybe a little bit more.

But if you have an AI tool or a platform that just twenty four seven, you know, regardless of holidays or weekends, is constantly monitoring, checking for, any inconsistencies in data where there might be a security risk or maybe someone up it in an API on a on a Friday night. Right? And then maybe the the mapping of the data is just off a little bit. And AI can step in, as kind of your builder copilot and then heal itself, well, you know, why you're enjoying the weekend. So I'm really excited to kinda see the movement and the progression of that and to see how platforms like Tray is kind of embracing that change, and just making my life a lot easier.

That's our goal. And I think on the on the flip side of that, you know, there's a lot about AI in the news, and there's a lot that, there's a lot that, you know, the the world's trying to figure out given the pace of change. What do you think some of the sort of misconceptions, could be or some of the ways in which, you know, we may be thinking about this the wrong way? Sure.

Well, I think everyone thinks that, like, AI is gonna be perfect day one.

It's not. There's large language models. Companies have to train it. You kinda get in what you, or you get out what you put in.

Right? So the more specific you are, the more that you know what the end goal is, the better. And then, in that same vein, people think that it's gonna take everyone's job. Right?

They think it's gonna take over everyone's job. It's gonna take over the world.

But at the end of the day, I saw a meme on LinkedIn a couple of weeks ago, and basically it said our clients still have to describe what they want.

Right? So, the majority of communication is actually nonverbal and nonwritten, and I think that's where us as humans have the advantage. We can kinda read in between the lines. We can ask provocative questions.

So AI is not gonna take our jobs, but I have learned that the more specific you are in putting in what you want the outcome to be, the more helpful it is, especially when building complex workflows, you know, like, on the Tray platform or just using something like ChatGPT.

Yeah.

I'd have to agree with you.

I think for me, you look at the pace at which technology gets adopted, you know, globally. And, you know, we look around and we feel as though everybody's now in the cloud, and everybody's adopted cloud technologies. And, you know, that kinda era is done. And yet if you look at the data, there are still so many organizations that are yet to fully adopt the cloud that is starting that journey that are kinda going through that digital transformation process. And I think for things like AI, it's an amazing assistive technology, and it, when harnessed correctly, it's gonna offer immense value. But it these things are not gonna come and fundamentally shift in the way that it's being publicized or be adopted in the way that it's being publicized without controls, without gating.

And they very much rely on, as you point out, you know, the descriptions, the definitions, the way in which they're driven in the first place. Secondary to that, it's also relative to the technology that it's connected to. Right?

AI obviously has a vast amount of knowledge that it's able to tap into, but it needs to be able to harness that by connecting to a tool or connecting to a service. So in Tray's world, connecting to our connectors or connecting to some of the way our infrastructure allows you to to process data enhances and changes the way that AI can be utilized.

So I think there's gonna be a great deal talked around governance and control and, us as humans, you know, having that stop button and being able to make the decision for it to continue and actually carry out an action on our behalf. But it's gonna make us look great. It's gonna help, you know, correct our spelling. It's gonna help us, process work faster. It's gonna help us carry out those tasks that would take us so long, previously, and I think that that, like, opens up an amazing opportunity for us.

You also mentioned LinkedIn there. I'd have to go and say for anybody watching this, it's so worth going following Stephen on LinkedIn.

His posts are extremely entertaining, but also very insightful. And a lot of the things that you create that you put out in the world, I think, would be so helpful for many other folks. So, a little plug there, but please go find Stephen on LinkedIn.

No. Thanks. You're gonna make me blush on this megacast. Don't do that to me.

But no. I appreciate it.

It's been fun. It's been fun over, like, the eighteen months or so seeing the progression of AI and then also just being firsthand and witnessing just how Tray is kind of embracing it up within their platform. So it's exciting to kinda see all these new features, and I'm thrilled to kinda dive deeper into it and then leverage it more day to day.

So I have one more question for you before we wrap up today. Sure.

Sure. We've talked a bit about some of the things that you've created, some of the ways in which iPaaS is gonna benefit from from AI, some of the misconceptions.

But this last question is kinda near and dear to my heart.

How do you think that AI is gonna enable iPaaS to sort of aid a broader user base? Right? It's, you know, it's no secret that when building Tray, our vision was always, how can we open this up to more and more people? How can we, you know, give them the power of an engineer? And it feels that, it feels to me like AI is gonna open that door. You know, in your world as you're dealing with all different types of customers and skill sets, how do you see that opportunity unfolding?

Sure. Yeah. So let's go back six, maybe seven years. You had, SOAP APIs, REST APIs, and the only way you're gonna integrate anything because there weren't massive, you know, ecosystems of applications like, you know, Salesforce has, maybe HubSpot has now.

You know, the only way to integrate something would be to have a highly technical person who knows API development. Right? And then we've transitioned into kind of the low code. Right?

That's kind of the next step to broadening the audience and making sure that it's just more accessible to the average person. So it's still, you know, fairly technical. But with the implementation of low code, a lot more folks can kind of drag and drop and kind of integrate things as they go. So AI is just the next evolution of that.

Right? It's making it even more accessible to a broader audience from pulling just basic report data. Maybe you have an intern who they don't know what filters to put in Salesforce. They don't know what report type to pick and choose, but they can put that request within a chat window.

Right? So, leveraging AI to kinda do that interpretation of what the person wants and, you know, pulling that report data, giving it to that intern, and then that intern is just looking like a superhero because, they got the data right for the executive or whoever it is that they report into.

Or maybe it's the middle line manager. Right? They are more advanced in that they know what they want. They can be more specific, and they can, you know, leverage AI to maybe build some basic workflows, integrate with LinkedIn, integrate with Google, integrate with Facebook, you know, whatever it might be, building those integrations.

Or, you know, me and my team, we're more advanced users. Right? So we might be leveraging it to, you know, modularly build on a much bigger workflow that's a lot more advanced. And we use it almost like maybe a copilot.

Right? We're doing kind of the stuff that that we wanna work on, but the stuff that we don't wanna work on, we can leverage AI. Maybe it's building out loops or branches or, you know, building out some of the logic that, you know, we can just send it, you know, a whole list of just do this. These are the parameters of what we want built, and it just builds it.

And then I am spending a lot less time, you know, clicking and zooming in and zooming out and mapping data when I could be spending that time doing something better.

Yeah. That that's the stuff that's really exciting to me, and I think you kinda hit the nail on the head earlier.

The description piece is so critical here, and what I think is kinda great about the way that low code has evolved is that, it's created more stakeholders within organizations. And as that's happened, that's meant that people have had to get better at describing the problem because you're now working with different teams. You're working within different departments.

It means that you're being more descriptive about what you're trying to achieve. That's gonna, you know, translate really nicely into this sort of conversational automation approach, this method of being able to describe what it is you want. And when you pair AI with a technology like Tray, which, you know, has a real advantage because of the way that we architect it, because of the way that we built out our connectors and provided everything with this sort of cloud first approach, it means that you can open the door to an entirely new audience. And as that happens, you know, for us, that will be realizing a vision in many ways, but it's so exciting to see, you know, what people go out and create, and that's the bit that makes us, you know, unique as individuals. So, it's been a real pleasure to be able to spend a bit of time with you today.

As I've said earlier, you know, if you're looking for an expert in digital transformation, or you wanna, you know, go and learn something on LinkedIn, please give Stephen a follow. He's fantastic for us to work with and a real delight within our Tray community. So thank you very much for your time today. Yeah.

Thanks, Rich. Thank you so much for having me. It's been great.

Cheers, Stephen. So that's a wrap. Thank you so much for tuning in today. Hopefully, you're as excited as we are to introduce Merlin into the world, let you see a little bit behind the scenes, and spend a bit of time with Stephen to understand his perspective on where this market is headed. We're so excited to be leading the forefront of the AI charge for automation. It's totally changing the game right in front of us, and really excited to catch up with you very, very soon. Thank you thank you all.