AI agents are quickly becoming the focus of large swathes of the tech industry, and the ServiceNow ecosystem is no exception.
As we recently reported, ServiceNow’s Q2 earnings revealed a significant boost for the ITSM giant, thanks in part due to artificial intelligence deals, with Chairman and CEO Bill McDermott saying: “The AI revolution is in full flight and knows no boundaries”.
Eddie Guo, Founder and Head of AI at Echelon, spoke to NowBen about his time building AI and machine learning products over the past decade, what he’s learned, and his predictions for the future.
Eddie’s Expertise
Eddie spoke to us about how he honed his craft during his time as a tech lead at Uber between 2016 and 2019, working on the “intersection of economics and machine learning” – particularly how they predict people’s willingness to pay for a trip.
He told NowBen: “Let’s say you’re going from your apartment to the Heathrow Airport, are you willing to pay £50 or are you willing to pay £100, right? Also balancing the supply and demand in real time in London to figure out what’s the price we should show Henry. We’ve been doing a lot of that for the past decade.”
He joined Moveworks in late 2019, helping build core AI capabilities, including enterprise-scale AI agents. It was here that Eddie met the two people who would become his fellow co-founders for his AI startup, Echelon.
Eddie said: “We all met at our previous company called Moveworks. We all joined early and really saw how the founders built up that company, and really learned a lot. I think that inspired us to be like, ‘Hey, we work together, we like working with each other, can we also do something similar?’
“I think that that’s also where we learned a lot about IT service management, about ServiceNow, we learned a lot over there, and also saw firsthand how they grew that company to become a $3B company, so that’s definitely a big inspiration for us.”
Eddie said that what really got him to build Echelon was speaking to nearly 100 people in the ServiceNow ecosystem, and he would constantly hear that a major pain point is development speed.
“We want to be able to deliver this AI agent that can work for folks, alongside your developers, and speed up your development,” he said.
Why Echelon?
Echelon utilizes AI agents to autonomously develop customers’ ServiceNow experience end-to-end.
When asked for an elevator pitch of the service, Eddie told us: “Echelon’s AI agents take in your business process stories and automatically discover, develop, document and test, and deliver the story to you.”
Providing more detail about the offering, he said that when you have a certain level of requirements and you have a clearly defined story of what needs to be done, you simply assign it to Echelon – which can integrate with your existing Agile board on ServiceNow, or tools like Jira or Asana.
He added: “You simply assign the work to Echelon, and Echelon will start working on that task… in a few minutes, Echelon would give you a workspace on where Echelon’s doing the work, and also give you notification when it’s done work, or when it needs more clarification.”
Essentially, the aim is for the user to be able to “fire and forget” – give Echelon a task and let it handle it. You’ll get a notification when it’s done.
Whereas tools like Copilot can answer your questions, it’s not actually doing the work, so there’s still a lot of juggling that needs doing.
Eddie said: “I think that tools like those are useful, but I think it’s a small productivity boost. I think we want it to be an AI agent or AI companion that actually works with you, do some of the work for you, so you’re focusing on more of the creative work.”
When talking about what their ‘ideal customer’ would look like, Eddie said it would be someone with a lot of backlog items, which current staff capacity is unable to catch up to.
ServiceNow Automated Test Framework (ATF) has also been a pain point, according to Eddie, who says a lot of people really want to take advantage of it, but during implementation, it’s typically the first thing they cut, because not having it “won’t break any deal right now” – but in a few months, while upgrading and making changes, you’ll realize you should have added some guardrails and tests to help catch different issues.
Eddie said: “That’s something we can help with because we helped our customers get from zero test to 60% of coverage in just a week or two, because we are able to ingest, understand your existing workflows and catalogs, build out and just come up with test cases that cover different paths off the workflow.”
What Does the Future Hold?
There’s all sorts of talk about how AI is going to change the world forever, and those of us who live in something of a tech sector bubble might do well to remain skeptical of hype surrounding one product or another.
But even taking our perhaps rose-tinted view of the topic into account, it does seem inarguable at this point that artificial intelligence at least has the potential to disrupt – in both a positive and negative sense – the way we currently operate in the workplace, and beyond.
This includes the possibility of human workers increasingly becoming redundant, when AI can do their jobs better – in the pessimist’s view; while the optimist says that repetitive, monotonous and time-consuming chores will be handled by AI, freeing up humans to pursue more meaningful and creative endeavors.
We asked Eddie, who has a decade’s worth of expertise in the subject, about how he sees these developing technologies affecting the future.
He said: “I think it’s going to be a hybrid world where a human is still going to be able to have that high-level picture of what needs to be done, what’s going on, connecting different pieces.
“But when you get to the point where the nature of the work is clear enough and AI agents are purposely built, when the task is at that stage… I think we should delegate those tasks to the AI agent, right? Otherwise, you’re going to fall behind.
“I think that if that’s the general trend the industry is going to, we’re already seeing that in customer support, we see companies, quite a few of them, doing really really well, they’re already resolving 80% of customer tickets and we see that in coding.”
He added that some businesses have implied they will not be expanding their workforces because they can “already do more” with their current numbers, thanks to AI help.
Why AI Agents?
We have seen artificial intelligence evolve in sophistication – and purpose – in recent years, from predictive, to generative, to the agentic.
It is sometimes said that the sky is the limit when it comes to agents, and this is supposedly a point in their favor – but so too might it be a drawback, when potential customers are left asking what precisely this limitless technology can do for them, today.
Eddie said: “I think it really depends on the nature of your work, what exactly is the work you’re delivering? Let’s say if it’s a self-driving car, right? I think the risk is very high. It took years for Waymo, Tesla, for some of the players to start just testing it out, and now they have a very limited zone where Waymo cars can operate. I think in our business it’s slightly different.
“So I think first of all we can clearly define, ‘Hey, these are all the actions this agent has available, let’s say, using a ServiceNow API – you just don’t have access to drop the database, to drop all the tables and remove all the users from the group, right?
“I think second of all is because it follows the developer workflow, you’re always doing things in a dev environment. What’s the worst thing that can happen? Everything has a version. You can always reset, right? It’s similar to a pretty safe sandbox.
“Third, you can always take control and override it. You still have a sort of fallback. I do see that, because this is an area that’s more internal facing, it’s expediting your development. It’s not directly interacting with your end user.”
Eddie says he has seen examples of people trying to jailbreak a support agent on the web, but it’s still “generally safe” because measures have already been taken to limit its capability.
Final Thoughts
While details can often seem fuzzy about how exactly AI is going to revolutionize the world, it seems apparent, to me, at least, that this change is going to happen.
Eddie’s bold decision to move into the world of agentic AI solutions will, in my opinion, prove to be a fruitful one, as more and more businesses see how these tools can effectively solve problems and streamline operations.
ServiceNow’s recent Q2 results certainly seem to show, as Bill McDermott said, that the AI revolution is in full flight.