A year with AI: you are still in charge!
What's your AI resilience score, and does it matter? A year of daily AI use brought less waiting and more judgement, but the accountability stays with the
Mission: BETTER! - What’s your AI resilience score, and does it matter?
Some days the debate about whether AI will take my job feels a bit like discussing star signs. This week I skipped the horoscope and gave RankedIn a go. It gave me 82 out of 100 for AI resilience, a laurel wreath and the title Orchestrator. I look delighted.
A day later, with a LinkedIn export three months newer, I scored 88! The first report saw an SAP sales leader. The second saw an SAP enterprise transformation executive. I happen to be both.
RankedIn.App
What neither report could tell you is what I’ve learned from using several AI tools every day for a year. Writing this article is a fair example.
I wanted to write about how AI has changed the way I work. One draft spent so long picking apart the career assessment that it started to read like a product review. I asked for less of that and for SAP to come back in where it belonged. The next draft took me into a discussion about who handles an exception at month-end close. It was a perfectly reasonable question in an SAP project, but quite a detour in an article about my year with AI.
I’d like to say a better prompt would have sorted that out on the first attempt. In practice, seeing a draft I disagreed with helped me work out what I wanted to say.
After more than 25 years in SAP and enterprise transformation, I bring experience to a conversation. I bring a few well-rehearsed answers, too. One of the best uses I’ve found for AI is to give it an argument I’m already comfortable making and ask, What would someone with equally good experience see differently? I can take that objection to another tool, give it the underlying material and see which parts of my argument still hold up. Occasionally I change my mind. Sometimes the exercise helps me explain why I haven’t.
The part I hadn’t expected was how much of that learning carries over. A question that improves one article earns a place in the next. A correction I find myself making twice becomes an instruction I can use from the start. And “this sounds convincing, but I wouldn’t say it” is a perfectly good reason to delete a paragraph. Business already has enough beautifully written paragraphs that say very little ;-).
I haven’t counted the hours saved.
What I notice most is how much less time I spend waiting.
In software, I’ve seen ideas sit for months, sometimes years, while we found the right mix of skills, budget and people with time to spare. Today, with several AI tools and a quiet afternoon, I can often get far enough to see whether an idea works. A first version can take days, a revision hours, and sometimes the answer to “what if we tried it this way?” is on the screen in seconds.
After more than 25 years in SAP and enterprise transformation, I think this may be the best time I’ve known to work in software. There’s a particular pleasure in having a question, closing the door and quietly making something of it while the question is still fresh.
There’s a risk in how much I enjoy that silence. An agent will work through an idea with me for as long as I like. After enough meetings spent hearing why something would be difficult, or why it couldn’t be done at all, that kind of conversation is easy to prefer. The uncomfortable part is admitting that sometimes the person raising the objection knows something my agents and I have missed. I still need to take what I’ve built back into that room. And if I ask people to act on it, I’m responsible for the outcome.
So, back to the laurel wreath. RankedIn gave me 82 one day and 88 the next. Both scores sit in the range it calls resilient, where it sees more opportunity than threat. That feels close to how I see my field today. It probably explains why I look so delighted in the picture.
Apparently this is how RankedIn.app sees me 😄
For me, AI is first of all freedom. Your mileage might differ.
From my desk to your programme
In the last issue, on ZEISS, I argued that the cost of producing facts is falling, and that prove before scale stops being a single checkpoint and becomes a cadence once facts get cheap. My year with AI is the small, personal version of that argument. It leaves a question for every leader running a transformation:
if a test takes an afternoon, why does your programme still decide once and defend it for years?
Two scores, one person
The gap between 82 and 88 is the most useful thing the two reports told me. Same person, same career, a newer export and a different reading. A score is an opinion about the evidence it was given.
That is why the career and AI resilience check on sapperment, Know your edge, gives ranges instead of a single number. Answer “not sure” and the range gets wider. The targets behind it are editorial judgement, set before anyone was scored, and the page says so. Nothing is uploaded; the answers stay in your browser. I would rather show people how uncertain an estimate is than hand out a precise-looking number.
What the year taught me about SAP work
In August I wrote on sapperment that AI is changing the shape of SAP work. The first thing it automates is the keystrokes. The value moves to judgement: process ownership, Clean Core discipline, data and integration, change and adoption. Being able to operate AI is table stakes; being able to judge its output is the skill. Robert Lienhard turns that into the question he brings to viva, sapperment’s live roundtable: which human contribution gains value once the keystrokes disappear?
A year at my own desk has made that concrete. The useful moments in the story above were rejecting a convincing paragraph, asking for the objection I did not want to hear and turning a repeated correction into a standing instruction. None of that is typing. All of it is deciding what is right and being willing to stand behind it.
Tobias Unger explains in his Transformation Office column on sapperment why the convincing paragraph is the one to watch. A language model produces coherent form, he writes, and the reader supplies the meaning. That is why “this sounds convincing, but I wouldn’t say it” is a test worth keeping: it is the moment I check whether the meaning is in the text or only in my head.
An SAP programme works the same way. An agent can draft a configuration, propose a mapping or write the first version of a functional spec. It cannot decide whether three business units really need three different exceptions, and it does not answer for the result.
The time AI gives back belongs to someone
I said I have not counted the hours. For me, the gain is less waiting. For an organisation, that answer is too loose.
In The capacity dividend, her Growth Mandate column on sapperment and the companion to The AI Investment Test, Nina Simosko asks the right question: when AI gives an organisation capacity back, who decides what happens to it? She separates task efficiency, usable capacity and realised value, and her line “An hour saved is not automatically an hour monetized” belongs in every AI business case.
My quiet afternoons are mine to spend. A finance team’s freed capacity is a management decision: cut cost, cover more ground or reinvest in growth. If nobody makes that decision, the dividend disappears into existing workloads and nobody can say where it went.
Still in charge, and still accountable
The paragraph I find hardest in my own story is the one about the room. An agent will go along with an idea for as long as I like. The colleague who raises the objection sometimes knows something the agents and I have missed.
In an enterprise, the same line runs between a copilot and an agent. A copilot suggests; an agent acts. Once software acts, the conversation moves from productivity to accountability. That is why the Executive Clarity pages on AI agents ask for every agent action to be explainable, auditable and reversible, and why governance for acting AI has to sit in the process design instead of in a quarterly review.
For the same viva episode, Rej Pathania put it in one sentence: “Let the machine propose; let an accountable person decide where being wrong is expensive.” Tobias Unger added the part about the room: technology exposes the seams in an organisation, and people still have to repair them.
That is what “you are still in charge” means to me.
The tools have made me faster and, on good days, better. They have not taken any of the responsibility off my desk.
If I ask people to act on what I built, the outcome is still mine.
Your turn
Freedom is my reading of the year. I would like to hear yours, and two questions in particular. Which decision in your programme would you test again if the test took an afternoon? And who would have to agree before you were allowed to change it?
If you want a reading of your own that is honest about its uncertainty, try Know your edge. It takes about five minutes:
https://www.sapperment.com/talent/know-your-edge/
Further reading on sapperment
The AI Investment Test (Nina Simosko, 5 November 2026): https://www.sapperment.com/insights/ai-investment-test/
The capacity dividend (Nina Simosko, 5 November 2026): https://www.sapperment.com/insights/capacity-dividend/
The limits of language and the power of context (Tobias Unger, 1 September 2026): https://www.sapperment.com/insights/limits-of-language/
viva Episode 1, The chain of control (Robert Lienhard, Rej Pathania, Tobias Unger and Philip Zhang, moderated by Bonnie D. Graham): https://www.sapperment.com/experts/viva/chain-of-control/
AI is changing the shape of SAP work (Andreas BORN, 15 August 2026): https://www.sapperment.com/talent/sap-careers/ai-shape-of-sap-work/
AI agents in ERP (Andreas BORN): https://www.sapperment.com/sap-executive-clarity/ai-agents-erp/
Embedded AI governance (Andreas BORN): https://www.sapperment.com/sap-executive-clarity/ai-governance/
Mission: BETTER! Independent executive perspectives on SAP Cloud ERP, Business AI and enterprise transformation. Grounded in experience, not marketing.
Independent publication. Not affiliated with or endorsed by SAP SE. SAP is a registered trademark of SAP SE.
First published on LinkedIn. Subscribe to Unfiltered: SAP Public Cloud to get new editions as they appear — or browse the sapperment library.