ServiceNow Knowledge 2026: Agentic AI propelled ‘from concept to a movement’

Observing customers

Garlanded as an expert in the CRM field, Chesire said the best way for him to differentiate valuable AI breakthroughs from the fads is to pay attention to customers’ needs.

These announcements are another step function in bringing AI in a thoughtful way, where we think it’s not ‘or’ – it’s ‘and’

“The North Star for us, and for myself, was, from the beginning … observing what customers were doing,” he said. “So, 11 years ago, when we were looking at the genesis for this [workflow], we saw customers customising on the platform … and we were very curious.

“They told us the systems they had at the time did a part of the process. They helped do interactions, but [our customers] still needed to stitch together [processes] to accomplish what they wanted.”

This helped inform one of the major selling points ServiceNow markets so effectively: an end-to-end experience where various apps seamlessly coexist to complete their complementary tasks.

“Even with AI, we still look for tangible business value,” added Chesire, providing an example of the issues he seeks to solve: “Customers say things like, ‘My team members are simply reading emails and scrawling through log files, and not actually spending time with their customers.’

“I just left a meeting with a Japanese customer who said, ‘We’re now getting much richer insights from the call summarisations and the Assist context than we ever did [before],’ because the machine is able to get a perfect summary of what the customer asked for.

“And the team members are much happier, because they get 60% more time to talk to customers, and not do all the work they were told they had to do before,” he added.

A quick headcount

But just as Big Tech encourages us to embrace our AI colleagues, stories abound of those same companies swiftly restructuring their workforces to maximise investment, while some of the unforeseen side-effects of the dizzying agentic revolution are yet to come into full view.

McDermott told the press at day one’s fireside chat there was no alternative to the pace of change, and noted ServiceNow’s AI training initiative as an option for those still unconvinced.

“Anxiety is understandable,” he said. “A lot of people actually really don’t understand all the nuances of AI: ‘It’s either good or it’s bad. It’s good because you can cut headcount: it’s bad because you can cut headcount.’

“But it’s so much more nuanced than that,” said McDermott. “We’re committed to our RiseUp programme, training three million people, at no cost, to get them initiated in the AI economy.

“If you think about the issue … if you take Germany, the UK, Japan and the US, by 2030, you’ll have a 50 million-person shortage in the tech community. … You have fertility rates in the global economy dropping, and you have a headcount of workers in the enterprise absolutely flatlining.

“What is going to improve productivity when, at best, you have the same amount of people doing the work?” he asked, before channelling the great American industrialist, Henry Ford: “If you always do what you always did, you always get what you always got.”

Nvidia partnership

Computer Weekly asked Chesire for his pick of the partnerships announced at Knowledge 2026, providing the perfect opportunity to summon the world’s first $5tn company.

“Nvidia is obviously a very exciting company, and a great partner,” he said. “In the sales realm, these are really powerful chips that they sell. What most people don’t realise is the chips go inside racks that go into chassis that have very specific requirements on communication cabling that goes in the back – power cabling, cooling requirements. These are massive systems.

“And so, the power of AI, and the deterministic core, to build those things out, is why they went from over five days to build one of these systems, to minutes,” said Chesire. “That is how [ServiceNow] makes a seller or even a customer who’s waiting for one of these systems get exactly what they need, better.”

With use cases like these, it becomes clear that highly ambitious tech executives are not best placed to pore over AI replacing the clerk who works to live, and how human agents such as these might be safeguarded going forward. 

Rather, Chesire’s insight into ServiceNow’s Nvidia contract speaks to a life’s passion for optimising workflows at the top table of the global economy, where minimising errors in the production line stands to save the customer unspeakable expense.

“A key part of what makes Nvidia different is its software that works with their chips,” he said. “Most people don’t realise that. Just like any other software, sometimes there are defects. But they’re hard to find. And so what we do on the AI agent side is understand and analyse the telemetry – what was going on; what was the workload; what was the unique condition? – to summarise and put it to a really sophisticated AI engineer, who really doesn’t want to spend their time going through log files, to say, ‘This was the specific thing that was going on at that time.’

“But then … there’s always the work we tell people they should do,” added Chesire. “One of the things we tell people is, ‘You should write a knowledge article about what you did to fix it.’ No one ever did it. So the AI agent also writes the knowledge article to say, ‘Here’s the specific model, the deep specificity on what was going on,’ and what fixed it, and it helps the next team member and the next AI agent do better.”