What Udemy is building with AI in Tines

For the security team at Udemy, AI in workflow automation provides an opportunity to unlock new time savings while keeping their organization secure, and protecting their online learning and teaching marketplace of 62 million users.

But like all good security teams, they don’t want to sacrifice data security or privacy. AI in Tines, which is secure and private by design, provides that all-important layer of control - data never leaves the region, travels online, is logged, or is used for training.

This makes the new AI features – automatic mode and the AI action – a great fit for Udemy’s security team.

I believe that we as a company at Udemy are looking at leveraging AI a lot more, in terms of our day-to-day work and in terms of what we bring to our customers. These kinds of features are very exciting for the team and very exciting for Udemy as a whole.

Kyle McGaley, Senior Security Engineer, Udemy

Kyle, a longtime Tines customer, has been looking forward to seeing AI in the platform since large language models became widely available. “I remember having a good discussion with (co-founder and CCO) Thomas Kinsella around AI a while back and how Tines would look at implementing it. So now to see it being implemented in such a well-oiled way is really nice. It’s been a common theme when I’ve worked with Tines - they deliver on a promise and implement it well.”

Kyle and his team participated in the Early Access Program for AI in Tines, and started seeing the benefits almost immediately. Within a couple of days, they had built several useful workflows.

Kyle started by using automatic mode to save time on event and alert management and response. “We have log sets that come from multiple sources that all have different shapes and patterns, and being able to describe a problem to get the same result across all of the different types of logs was really helpful. That’s one big time saver that I’ve found, being able to format things in an easy way, regardless of the input.”

With the AI action, he's formatted the custom emails the workflows will send to users for different types of alerts. “Each type of alert is going to have slightly different email content and layout. So it’s great to run it through the AI in Tines say, ‘Hey, we have this template, make these changes and show me the output.’ Most of the time, within an iteration or two, the output is what you want. That definitely saves us a lot of time.”

He adds, "I really like the ease of popping a prompt in and solving a problem faster. Automatic mode is a game changer, because there’s always a thought process that goes into how you fit an event transform into your workflow and what it needs to do. "

It does the heavy lifting for you, which is nice, especially when it comes to Lambda functions and the trickier parts of event transforms.

Kyle McGaley, Senior Security Engineer, Udemy

Before Tines, Kyle was already actively using LLMs to assist with building his workflows, for things like email communications and creating Slack blocks. “Using the AI action, you can do all those things within Tines. It saves you having to pay for a subscription to another LLM.”

In terms of what you can build with it, the limit is your imagination. You will save a boatload of time when it comes to building out emails, and building out responses to users.

Kyle McGaley, Senior Security Engineer, Udemy

The usability of the features is key for Kyle. “You can just click a button, type a sentence, and instead of having to build HTML, and format it, the AI action does it for you. And if you don't like how it does it, you can just adjust it and it will fix it. Those kinds of tasks save a massive amount of time in the long run, and it will build up over time, the more you use it.” 

Kyle’s team is already thinking of new use cases for the AI-powered capabilities.

I could see the potential of Slackbots, incident response bots, basically the streamlining of the security response and instant response systems.

Kyle McGaley, Senior Security Engineer, Udemy