The Most Expensive Free Code You'll Ever Write: AI, ServiceNow, and the Cost of Not Knowing the Platform

I'm a bad technologist. I'm slow to move to the latest and greatest, and I'm skeptical of anything that claims to make things more efficient. When direct deposit first came out, I did not sign up for it. I didn't trust that my money would arrive on time, and I wanted to see for myself that my check was correct. Today? Please don't hand me a check or cash.

I've had some of the same hesitations about AI. Let's set aside the people who use it maliciously or to create false narratives, and the environmental impact (that's a whole other soapbox). Let's also set aside how tired I am of hearing about AI.

I was a slow adopter. Using AI felt almost like cheating. It seemed to discount the experience and knowledge I'd built over 15+ years in the trenches: the late nights writing papers, sometimes at the 11th hour, and the late-night go-lives where the real relationships, friendships, and lessons happen. It felt like taking the easy way out, and like it might take away the parts of my job I loved most.

As I've slowly embraced AI, I've learned a few things worth sharing, in case you're having the same thoughts or wondering, "How can I use AI to build out my ServiceNow platform?" (or do any other job). What I've found is that AI doesn't replace expertise. It amplifies it, and it exposes the lack of it. I recently watched a customer learn this the hard way, and pay for it twice.

AI is NOT magic

Magic? No. Impressive? Yes. But getting good output from AI depends almost entirely on how well you write the prompt. Without knowledge, wisdom, and experience, you won't know what to ask for to get the outcome you need.

I was experimenting with Now Assist to see what it could code for me. It did a good job, and I thought, "Oh, I see why people think consulting companies and partners won't be needed anymore." Then I looked back at my prompt. I had written it the way only someone with years of experience, trial and error, and platform knowledge would. I knew exactly what to write to get the result I wanted. I knew the output followed best practice. I knew it was scalable and maintainable. And I knew to click "No, I don't want you to go against best practice."

Does that mean someone without that background can't get their platform where it needs to be?

Not necessarily, but it will be a lot harder. They may miss best practices or end up with solutions that don't scale. They may miss features that are built into the platform, not for lack of interest, but because they don't have the formal training and years of hands-on experience.

And those gaps don't show up on day one. They show up months later as technical debt, customizations that break during upgrades, rework, and a platform nobody wants to touch. AI will happily build you something that works today. It won't tell you it's going to hurt you tomorrow.

As I kept working with Now Assist, it did speed up some of the more tedious work, but only because I knew how to ask. I knew the solution and explained it clearly. I also needed fewer prompts and fewer assists. The tokenomics of my session (the cost of all those prompts and AI credits) were far lower than they would be for someone who has never worked in the platform. That person will need more assists, more iterations, and more time to get the system where it belongs.

AI is NOT a mind reader

Sometimes I think I've written a great prompt. I've set the persona, defined the problem, and laid out the solution. Then the AI does something completely out of left field, and I'm left wondering how it missed what I wanted. Something got lost between my brain and the keyboard. That happens without AI too, but it's especially frustrating to watch it run, only for it to come back completely wrong or nowhere near what you expected.

Knowing when the output is wrong, and why, is exactly the experience AI can't supply.

AI is NOT replacing you

As SVP of Delivery, I hear a lot of questions, concerns, and fear about AI taking my team's jobs. Will parts of our work become commoditized? Sure. That happens with every technology shift. Will AI replace my team's expertise and knowledge? No. AI is only as good as the person giving it input, reviewing what it produces, and confirming it's right.

We're also seeing that AI, in inexperienced hands, can end up costing more than seasoned professionals would have. We recently worked with a customer who used an AI tool to build out Change Management. Without realizing it, they modified underlying code that Change Management depends on. The process broke so badly that everything had to be reverted. They paid for that work twice: once in the time and AI spend on the original build, and again when they brought us in to fix it.

The real advantage isn't AI or experience. It's both. An experienced team using AI moves faster and costs less than either one alone. Don't get me wrong, I enjoy a good AI cat video as much as the next person. But the experience and knowledge you bring to AI are what set you apart.

The bottom line

AI is a tool in our toolkit, and that's how I've embraced it. It helps me get up to speed quickly on a new project or escalation. It helps me clean up emails and PowerPoint decks. It has also made our Developers, Project Managers, and Business Process Consultants more efficient. That lets us pass savings on to our customers and spend more time on the consulting and human connection that produce the best outcomes on the platform.

Can you build out your ServiceNow platform by having someone without platform experience prompt AI? Sure. Will you be calling a partner when what you built won't scale, breaks on upgrade, or can't be maintained? Probably. AI has changed how fast we can build, but it hasn't changed what it takes to build well. A seasoned partner using AI will get you to value faster, at lower total cost, and with a platform that still works for you years from now.

Use AI. Just make sure someone who knows the platform is the one holding the prompt.