I keep meeting owners who have delegated their entire ai business strategy to whoever runs IT, the way an earlier generation delegated "the internet" to the webmaster in 2003. It did not work then, and it will not work now. AI is not a system you install. It is a decision about how your business does work, and that decision belongs with the person who owns the operating model, not the person who owns the servers.
The confusion is understandable. AI arrives dressed as a technology purchase: a chatbot, a model subscription, an API key. So it gets routed to the technical team by default, the same way a new accounting package would be. But an accounting package does not change how fast you can respond to a customer complaint, how many people you need to process an invoice, or what your cost structure looks like in two years. AI does all three, which makes it an owner-level lever, not a procurement decision.
I have watched this misrouting cost companies a year of hesitation while competitors who treated it correctly moved faster and cheaper.
What Changes When AI Sits With IT Instead of the Owner
When AI strategy lives entirely inside the IT department, three things happen predictably:
- It gets evaluated on technical merit only. Does the model work, is the latency acceptable, is the integration clean. Nobody asks whether it changes headcount needs, service speed, or unit economics, because that is not IT's job to ask.
- It stays a pilot forever. Pilots are safe for IT to own. Committing to a new operating model, cutting a support team's headcount growth, or promising customers faster response times is not something a technical lead can authorize alone.
- The budget stays trivial. A tool budget gets a few million rupiah a month. A strategic bet on cost structure gets board-level attention. AI initiatives routed through IT rarely escape the tool budget, which caps how much they can actually do.
None of this is a criticism of IT teams. It is a criticism of where the decision sits. The technical evaluation belongs with them. The strategic call does not.
The Owner-Level Questions IT Cannot Answer Alone
Ask these instead of "which AI tool should we use":
- Where does our cost structure bend if response time drops from hours to minutes? This is a staffing and service-level question, not a technical one.
- What decisions are we currently paying people to make that a model could make consistently, at the same quality, for less? This requires knowing the business, not the API.
- Which of our competitors have already made this move, and what does their unit economics look like now? Competitive positioning is an owner's job.
- If this works, what does our team look like in eighteen months? That is a workforce planning decision with real human consequences, and it cannot be delegated to a systems administrator.
These are the same category of question an owner asks about opening a new location or changing a pricing model. AI deserves the same seat at the table.
A Framework for Owning It Without Learning to Code
You do not need to understand transformer architecture to make this decision well. You need to understand your own cost structure and where the friction sits. A practical way to start:
| Step | Owner's job | IT's job |
|---|---|---|
| 1. Identify friction | Name the three slowest, most expensive, most error-prone processes in the business | Confirm what data exists to support automating them |
| 2. Size the opportunity | Estimate the hours or errors it would take out per month | Estimate technical feasibility and rough cost |
| 3. Decide the bet | Choose one process to commit budget and a real deadline to | Build or configure the chosen solution |
| 4. Set the bar | Define what "working" means in business terms, not uptime terms | Report against that bar, not a technical dashboard |
| 5. Review quarterly | Decide whether to expand, adjust, or kill it | Maintain and flag technical drift |
This mirrors the same discipline behind the one page digital strategy every SME can write: the owner sets direction in business language, the technical team executes against it.
Why This Matters More in 2024 Than It Did a Year Ago
Model prices have fallen sharply through this year, and capability has improved fast enough that what required a research team eighteen months ago is now a weekend integration. That combination means the businesses that treat AI as a strategic lever, not a line item, are the ones compounding an advantage right now. The gap between "we experimented with a chatbot" and "we restructured how customer service works, at lower cost, with faster response" is not a technical gap. It is a decision-ownership gap.
If you have already read about what AI-native operations actually means for a business, this is the piece that precedes it: the operating model decision has to come from the owner before the operational redesign can happen at all.
Takeaway
Do not ask your IT lead which AI tool to buy. Ask yourself where your business is slowest, most expensive, and most error-prone, then bring that question to your technical team as a brief, not a request for a vendor list. AI strategy is a cost-structure and service-speed decision, and those decisions have always belonged to the owner. Keep them there.