In this demo, Itential Solutions Engineer Joksan Flores tackles a challenge that rarely gets discussed in network automation: language barriers. While most technology defaults to English, teams operating across Latin America, China, the Middle East, and beyond are forced to work in their second or third language – whether configuring network devices, documenting tickets, or writing automation logic.
FlowAI changes that. Using FlowAI’s agent builder, Joksan creates FlowAgent entirely in Spanish that automatically documents ServiceNow change requests and delivers bilingual email reports in minutes – no code required.
Network engineers and IT professionals worldwide can now create agents in their native language – Spanish, Mandarin, Arabic, Portuguese, or any other language.
Teams in Latin America, Asia, Middle East, and Europe can build network automation and IT workflows without English-language barriers. Domain experts work in the language they think in, not the language technology traditionally required.
A single FlowAgent can generate documentation, notifications, and reports in multiple languages simultaneously. ServiceNow tickets, email updates, asset descriptions, and system documentation can reach global teams in their preferred language.
Joksan Flores • 00:00
Hi, everybody. Today we are in FlowAI and I thought I’d record a quick thing that I’ve been thinking about for a little while. Recently I was having a discussion with John Capobianco about the different things that agents can help us with other than the standard stuff, coding and reasoning through various tasks and so forth. And one of the things that came up was language barriers. I am a Spanish native Spanish speaker, and a lot of times it’s kind of difficult to translate concepts for a lot of people that are from other countries, right? Most technology is created with English in mind 1st . So you think about somebody in China or in Latin America or in the Middle East has to learn English so that they can execute commands on Linux servers or execute or configure CLI on a router or a switch or anything else, or perhaps even to document tickets, right, when they work for international companies.
Joksan Flores • 01:01
So today I thought I’d just take a very simple use case, just like the latter I mentioned there, and talk about ticket documentation and how language models can help us doing this with agents and leveraging the integration that we provide through our FlowAI platform. So I’m going to go ahead and start an agent. I’m going to create an agent fully in Spanish, but I will actually explain it in English as well, just to kind of capture what it’s doing. So I’m calling the agent change documentation, and I’m going to paste the prompt so that we don’t have to type in the whole thing here. And then I’ll walk through it in a fair bit of detail. And I also selected a ticket from our ServiceNow instance. So what we’re going to do is we’re going to create an agent that provides change documentation, but we’re going to make it bilingual.
Joksan Flores • 01:49
So, now what we’re going to do is, I, as the persona that’s creating that agent, I am creating the agent in Spanish because that might be the language that I’m more comfortable in, but I am also asking the agent to create a summary of a ticket in English. So, here to translate really quick over the system prompt is: your job is to document, change requests, and send reports via email. Obtain information about the ticket provided and send an email with the key details about the ticket in English and Spanish. Format should be or must be in HTML with two brows or a table format. First, we’re going to include the summary of the ticket in Spanish, then we’re going to include the summary of the ticket in English, and the summaries should be of 20 lines or less. And then I also selected a ticket from our ServiceNow instance that just has a support turnover ticket. This could be applicable to anything.
Joksan Flores • 02:49
And just start thinking about the possibilities, right? All the things you could do. You could create entire agents that provision environments using your native language rather than having to do it in English. So I here said I document this ticket. So let’s go ahead and go next. I already have a project that’s called Agent Tool Belt. We use it quite a bit, and essentially is a Swiss Army knife of agent tools.
Joksan Flores • 03:16
It includes tools for create change requests, sending Slack notifications, sending emails, obtaining information from change requests, etc. So it has a handful of workflows, a few workflows that are useful for these. So these are my advanced tools that do certain logic. In my case, I actually have a workflow that does a fair bit of, you know, three or four queries into ServiceNow to obtain different types of information from the ticket. So that becomes my one tool that my agent will use for this. So I’m going to go ahead and configure and go to next here. I am not changing my LLM provider parameters or anything like that.
Joksan Flores • 03:53
I’m going to go ahead and create my agent. I will go and actually verify that everything looks good from my creation standpoint. My prompt is good, my project looks good, and my ticket, my user prompt is good. So I’m going to go ahead and launch that. And we’ll wait for the output. Okay, so my agent has finished. I’m going to go ahead and click inside of it.
Joksan Flores • 04:18
It ran, it took 25 seconds. And I’m curious to work through the trace, right? I want to work through that stack trace here or the trace of tools. We always, as we know, in FlowAI, we get the 1st two are prompts. We get, okay, the agent’s actually reasoning in Spanish entirely. Wow, that’s pretty crazy. So it says, I’m going to obtain the information from the ticket, and then I’m going to send an email with the details in English and Spanish.
Joksan Flores • 04:44
It went ahead and obtained a bunch of information. Like I said, I designed a workflow. So for workflows, you can start thinking about I could use an 80 call to ServiceNow , but service now’s API doesn’t quite have one endpoint that reports on all the attributes of a ticket. So I actually had to execute a handful of calls. So I created a tool, I created a workflow, which becomes a super advanced tool in my case that extracts a lot of information from the ticket, namely all the stuff, right? The change request ID, the current state, some of the who created it, the URL. Anyways, there’s a lot of data there from that change request, and that’s just all obtained via workflow, which becomes my super useful tool there.
Joksan Flores • 05:27
Now it says I’m going to send an email with the details in HTML format with a couple rows or in the table in Spanish and in English. Then it built the HTML, it sent it, and then we have a summary. It says, I have documented the ticket. I have sent an email with the key details for the change request. This is the summary of the ticket. Tickets closed. It was a port turnoff service from IAP.
Joksan Flores • 05:53
This is the device. And then the solicitor, which happens to be me. So the email has been sent. First row is in Spanish, 2nd row in English. This is all interchangeable, and we could do multiple languages. So just imagine if you have a team that spans the world and they have multiple languages that you got to deal with. And perhaps, you know, English is not the 1st language.
Joksan Flores • 06:17
So that might be a challenge or a barrier. Now you can actually document your tickets in this way. So let’s go and look at my email. So I actually got an email. Here we go. Ticket documentation for the port turnoff service, summary of the ticket in Spanish, ticket number rather, the state, it was successful, change, all the attributes for it. It was a normal change.
Joksan Flores • 06:45
Priority four, impact three, moderate risk. People that submitted it, who was assigned to, when did it open, when did it close, and the essential platform that was involved with it. And then in change summary in English. So now we got the documentation of this ticket in both in Spanish and in English. Now, I could have picked any other use case for this. Just think about the possibilities. Now, if you have teams in different countries, they can design their own agents in their own native languages, which makes it super strong from a prompting standpoint instead of having to use English as a 2nd , 3rd language.
Joksan Flores • 07:25
That might not be the 1st choice for that person. And they could also enact things like this when they can have translation or multilingual. Ticket updates. You could do this in the tickets themselves, in the descriptions of certain assets, in the emails that people are sent, etc. Think about the possibilities. This is something that’s a super awesome capability that we might be able to just take advantage of today by the use of Agentic AI. And in this case, I am, of course, using our FlowAI solution to do that extremely quickly.
Joksan Flores • 08:00
Hopefully, this is enjoyable, and thanks for tuning in.
See how Itential connects AI reasoning to governed execution across your entire infrastructure.