Sponsored by
Hey {{first_name | CS Pro}},
Most of us are still selling the value of AI on efficiency, and your customer's CFO stopped caring about that a while ago. The question landing on their desk now is simpler and harder, which is what did all this AI spend actually save us, in money, this quarter.
The teams getting AI adopted are the ones who can benchmark a process before and after, put real guardrails around what an agent is allowed to do on its own, and put a number on it that a finance lead will accept. Everyone else is stuck saying the word efficiency and hoping it lands.
This week on the podcast I sat down with Neil Sparkes (or Sparky, as everyone calls him), AVP of the Customer Excellence Group at ServiceNow for UK and Ireland.
His job is to move agentic AI out of the experimental phase and into live production inside large, heavily regulated enterprises.
I talk to him on the podcast on how heavily regulated enterprises are getting their customers to adopt AI right now.
But first, today’s sponsor
↓
Save 10+ Hours a Week With 37 Claude Prompts
Every manager faces the same situations before lunch: a message to land, a meeting to run, a hiring call, a report due. The AI Report built 37 Claude prompts for exactly those moments, organised by the situations every manager faces.
Copy the prompt, fill the brackets, run it in Claude, and get back 10+ hours a week. Oh, and it's free.
All you have to do is subscribe to The AI Report, a 5-minute daily AI brief read by 400,000+ business leaders at IBM, AWS and Microsoft, and the full prompt pack lands in your welcome email. The newsletter and the prompts, both free. Subscribe and grab both
Efficiency is not a value story
Neil learned this back in his infrastructure days, it is very hard to turn efficiency gains into hard metrics, because unless you are willing to reduce headcount or take a piece of software off the balance sheet, most finance leaders are not that interested.
His line was blunt.
If you tell a CFO you are going to make the team 20% more efficient, they will ask if that means you are cutting 20% of the team. And if the answer is no, you now have to prove your team is doing 20% more work.
Most of us cannot prove this.
What ServiceNow does instead is benchmark. They have years of workflow data sitting in their customers' systems, so they can show how long a process took before, and what it looks like once agents are layered on top.
Neil also said something I needed to hear. Do not get so bogged down arguing about the data that you never start. If you claim ten minutes and the customer says it is really five, that is still a 50 percent saving. Directionally right is enough to begin.
Treat your agents like new hires
This was my favourite part of the conversation. Neil's framing is that an AI agent is a person in your organisation, except you can hire a thousand of them in an hour.
Sounds great until you think it through. They still need access to systems. They need data. They need something that looks a lot like onboarding. They need to talk to the thousands of other agents out there, so who is getting them talking to each other? And just like a human, one of them can go rogue, touch a system it should not, or skip a corporate process.
So where do you start? Somewhere low risk, where the damage is small if it goes wrong. For ServiceNow, incident triage was the obvious one. An agent reads the ticket, adds context, spots that several similar tickets all point at the same network switch, talks to the monitoring agent, and connects the dots in milliseconds.
The human line gets drawn with guardrails. The agent executes up to a point, then comes back and says here is what I found, here are three options, which do you want. Neil compared it to apprentices straight out of school. You tell them exactly how far they can go.
The gap nobody notices they are in
Neil calls it the AI divide, and he says it is widening faster than anything he saw with cloud. The fast adopters accept some risk, get going, and pull away. The laggards are not falling behind on purpose. In his words, they do not even see it.
There is a flip side he was honest about too. He is drowning in information, because his team produces more with AI, and then he uses AI to summarise what they sent him. He called it AI laziness, or not marking your own homework. I call it outsourcing your brain. Somebody human is still at the end of that chain, and no amount of agents fixes that.
This week's challenge
Pick one customer this week and ask them the benchmark question. How long does this process take you today, and what does it cost you to run it? Then agree what you will measure against in 90 days. Do not build the perfect model first. Neil's whole point is that customers will tell you where they stand if you just ask, and once you have a starting number, every value conversation after that gets easier.
If your team needs help turning that number into a story a CFO will sign off on, that is the work Neil brought me in to do with his UKI team. Details at thecustomersuccesspro.com/team-event.
Have a listen to the full episode on Spotify, Apple Podcasts, or YouTube.
I hope you enjoyed this week’s newsletter.
If you have any questions or suggestions, please feel free to contact us.
Cheers to your CS success,
Anika
