# AI-First or AI-Enabled? Why the Name of Your AI Strategy Comes Second > In episode 6 of Prompt & Proper, Steffen Ochsenreither and I walk through the terminology zoo: AI-First, AI-Native, AI-Enabled, AI-Ready, AI-Shaped, AI-Augmented. Why AI-First wins at C-level and still sends the wrong signal, why AI-Augmented fits software development best, and why companies want the label in the first place. - Date: 2026-08-26 - Source: https://romanoroth.com/en/podcast/e06-ai-first-oder-ai-enabled/ - Tags: AI, Podcast, Prompt & Proper, AI Strategy, Transformation, Leadership As Chief AI Officer I am supposed to put an AI strategy in place, and the headline alone turns into work. AI-First, AI-Native, AI-Enabled, AI-Ready, AI-Shaped, AI-Augmented. In episode 6 of **[Prompt & Proper](https://promptandproper.ai/)**, Steffen Ochsenreither from Swiss Post and I go through the terms and end up somewhere the question does not suggest. The episode is in German, and this post captures the core of the conversation.
## AI-First Has a Sequencing Problem Steffen opens with the term that has won, and considers it the hardest one to defend. His argument: AI-First prioritises technology over structural change. The focus sits on technology, on hype and on trend, and hypes pass. What remains is the structure, and that is exactly what the term leaves out. My objection comes from the other side. We are having this discussion internally right now, my CEO wants to go that way: > "I am AI-first. So what does that mean? Is the human second?" The intended meaning is a different one. AI-First does not mean AI always comes first. It means that when we hit a challenge, we first consider whether we can solve it with AI. If yes, we do. If no, we solve it the classic way. And there is the catch. I have to explain the term for it to mean what it is supposed to mean. ## AI Glitter on a Broken Process Steffen sees a concrete danger growing out of that. If I check every problem for whether AI can solve it, I end up automating bad processes simply because it is possible. The workflow itself never gets touched. I consider that the biggest risk at this stage. You take the existing process as it stands and throw some AI glitter on top. You have gained nothing. Worse: a model is not deterministic, your old process may well have been. You can actively make the process worse. Steffen's addition covers the organisational side. Poor quality scales exponentially, and a new layer ends up sitting on top of the unfixed system. Organisations are immune to change anyway, they absorb it. If you change nothing about the organisation and only build AI on top, you get burnt out teams and resistance. Employees reasonably ask why they should do it differently now. It runs more automatically, not better. ## AI-Enabled and AI-Ready Start at the Foundation I brought up AI-Native, by analogy to the digital native. Steffen finds it better, because it at least implies people can handle the technology. He would still have to explain it, and his mother would still look at him blankly. On substance his preference is AI-Enabled or AI-Ready, because they start at the foundation. In most companies that foundation is poor: no data, bad data quality, no guardrails. Work on that, put it in the strategy, and you can build use cases on top. My problem with it is the marketing: > We are a bit AI-ready sounds like Intel-ready. Those stickers used to sit on laptops, and the machines were still bad. Steffen concedes it does not look good on a PowerPoint slide. It does point at where the actual work sits. And the topics there are unglamorous: data quality, availability, digital ethics. Which guardrails do I use, when, how? How do I calculate a business case with AI? Basics, and without them it gets hard. Digital ethics is the one Steffen sees addressed least often in companies. ## Maturity Decides, Not Taste Steffen does not settle on a single term at the end. For him the right one depends on the maturity of the company. AI-First is available to anyone who is a tech company, already has the foundation and can handle whatever comes. Very few sit at that frontier, in his view perhaps the large frontier labs. For everyone else: > Ready does not only mean knowing what exists. It takes the structural means and the will to change. Two topics that have nothing to do with technology. The technology comes after. Anyone doing AI-First inverts that order. Start-ups are his counter-case. A start-up that is not AI-ready today has a problem depending on the industry, simply because no historically grown structures stand in the way. ## AI-Augmented, and Why It Fits Software Development The term we both land on is AI-Augmented. Steffen puts it like this: > "It is not that I let AI do it, it is that I let somebody do it with AI." That brings people along instead of replacing them. For software development I consider it the most accurate term there is, because the human still has to stay in the loop. Models hallucinate, so the output has to be checked. The checking can well be done by a machine again, and not necessarily a generative one. A plain CI/CD pipeline with rule-based checks does the job, and generative agents then handle the fixes. Only the look at the pull request and the judgement of whether this is good stays with the human. Steffen's point on that: a process mapped that way already matches an AI-first company. Except that you did everything else in the background first. Trained people, provided tools, analysed the workflow, developed best practices. You need that maturity, otherwise you cannot introduce it at all. Why software development is furthest along, Steffen explains through process quality. The development process is well described today and looks very similar across many companies, so it is easy to compare and easy to improve. Individual business processes are often not even mapped, and what is not mapped cannot be automated. I would add that even in software development we look hard at the implementation. The part before it, ideation, budgeting and requirements engineering, and the whole operational part behind it are still weak in most companies. ## Why Companies Want the Label at All The real question is what the term is needed for. Companies expect higher productivity, higher margin or more revenue from AI, and they want to anchor that strategically. My counter-question: could you not do that directly through suitable strategic goals and OKRs, without a label? Steffen's answer is honest and explains the whole terminology zoo. Yes, you could, it just achieves less. Most companies have been digitalising for five to ten years. Writing "we want to keep digitalising" into your strategy moves nobody. Writing "we are doing something with AI" gets you considerably more buy-in, more attention and more budget. > AI does not solve problems that were not there before. It solves them differently. Data quality and data availability were open issues already. What is new is a vehicle those topics can ride on. Steffen explicitly does not want to talk that down: if a problem that sat in the shadows finally gets addressed, the term has done its job. ## The Take-Away My favourite from the episode is AI-Augmented, Steffen's is AI-Enabled, and the difference is smaller than the discussion suggests. Both of us would have to explain either one. What actually counts, Steffen put this way: > "When I lift the curtain and find it is very thin behind there, I stop caring what the name is." If you need an ambitious goal because you are wearing the C-level hat, take AI-First. Then put real thought into the explanation, and it will not be two sentences. It names which operational problem gets better, which process gets optimised, where the productivity comes from. The more concrete that gets, the less the term is the problem. If you enjoy the episode, subscribe to **Prompt & Proper** on [Spotify](https://open.spotify.com/episode/0PLE404YeJIPvxtDgGzKlO), [Apple Podcasts](https://podcasts.apple.com/us/podcast/prompt-proper/id6792786850), [Amazon Music](https://music.amazon.com/podcasts/b8a56367-a595-4865-8e73-97c706fa4743) or [YouTube](https://youtu.be/vVxG_eOzloo) and tell us which term is your favourite. All episodes: [promptandproper.ai](https://promptandproper.ai/).