AI can widen the search for options, keep the model of the business alive, and stress-test recommendations before resources are committed [1]. If everyone has access to similar models, advantage shifts to proprietary data, distinctive decision processes, and speed of learning. The next step is a decision operating system: Signals → Options → Scenarios → Challenge → Decision → Execution → Learning.
Were those the best markets — or just the ones we had time to research?
Imagine a leadership team evaluating entry into a new market. They have a deck with three candidate countries, financial projections, and a risk matrix. After several meetings, they pick one.
But there's an uncomfortable question: were those really the three best markets, or simply the three the team had time to research?
For decades, much of business strategy has been shaped by a constraint we rarely make explicit: people's limited capacity to gather information, analyze alternatives, and process complex scenarios. Classic tools — SWOT analysis, portfolio matrices, annual plans — aren't useless. They were designed to make decisions under those limitations.
Artificial intelligence is changing that starting point.
In an article published in Harvard Business Review in September–October 2026, Professor Felipe A. Csaszar argues that AI can widen the search for strategic options, enrich the representation of the competitive environment, and subject recommendations to more systematic challenge [1]. The implication goes far beyond producing reports faster.
The question is no longer how much work AI can do for us, but which decisions we can make better because of it.
From automation to decision intelligence
Today it's easy to find companies using AI to draft emails, summarize meetings, produce presentations, or automate responses. These are useful applications, but in many cases they leave the way the organization works untouched.
A company can save hours preparing its leadership committee and still make decisions with the same incomplete data, the same limited alternatives, and the same biases.
The strategic opportunity is different: redesigning the decision process, from framing the problem to evaluating results. We can understand this shift through three capabilities.
1. Explore: widen the universe of alternatives
Teams usually compare a small number of options. Sometimes because others don't exist; often because researching them takes too long.
AI can help identify acquisition candidates, product combinations, customer segments, pricing models, geographic markets, and possible competitive responses at a scale that used to be impractical.
Consider a B2B services firm looking to expand in Latin America. Instead of starting with the two countries its team knows best, it could build a broad list of opportunities, compare them by demand, competition, regulation, entry channels, and internal capabilities, and dig deeper only into the most promising ones.
That doesn't mean an AI should choose the market. It means leadership can decide from a better-informed set of possibilities.
Discipline still matters: a list of a hundred invented options is worse than a list of five verified ones. Breadth only creates value when there are criteria, reliable sources, and mechanisms to rule out what isn't viable.
2. Understand: from static diagnoses to living models
Most strategy exercises capture a snapshot of the business: a budget, a competitive map, a customer analysis, a sales forecast. By the time the result is presented, part of the information has already changed.
AI makes it easier to combine structured information — transactions, commercial metrics, inventory — with messier signals such as interviews, industry reports, customer feedback, or competitor moves.
A sales team could combine CRM data, customer behavior, margins, and market signals to spot opportunities a conventional segmentation doesn't reveal. An investment firm could connect operating data from its portfolio companies with external alerts to detect changes before its quarterly review.
This isn't about replacing one dashboard with a fancier one. It's about building a representation of the business that updates itself, gets questioned, and helps anticipate decisions.
3. Challenge: introduce useful disagreement before committing resources
Strategic decisions rarely fail for lack of slides. They fail because the right assumptions weren't questioned, because hierarchy silenced objections, or because the team fell in love with a solution too early.
AI systems can be used to run an adversarial review: one agent develops the thesis, another identifies weaknesses, a third anticipates the competitor's reaction, and another examines regulatory or operational consequences.
For example, before approving an acquisition, a committee could require the analysis to include the main reasons not to buy the company: customer concentration, margin fragility, integration risks, downside scenarios, and assumptions that haven't been validated yet.
An AI simulation is not the same as interviewing customers or getting legal advice. Its value lies in surfacing questions and contradictions that deserve real investigation.
If everyone has access to the same AI, where's the advantage?
If two competitors can use similar models, access to an advanced tool stops being, on its own, a sustainable differentiator. Advantage tends to shift toward three assets that are much harder to copy:
- Proprietary data and knowledge. Commercial history, operational information, project learnings, customer relationships, and knowledge that was never well documented. Quality, relevance, and authorized use matter more than volume.
- Distinctive decision processes. Connecting a model to company documents isn't enough. It has to be built into how investments are prioritized, initiatives approved, resources allocated, and hypotheses reviewed.
- Speed of learning. The ability to execute a decision, observe results, and adjust before competitors do. Not simply deciding faster, but shortening the time between a hypothesis and the evidence that confirms or challenges it.
A field study on innovation at Procter & Gamble offers an additional signal: researchers found that AI use could improve performance on product development tasks and help integrate commercial and technical perspectives that were normally separated by function [2]. It's evidence from a specific application, not a guarantee of equivalent results in every organization.
The lesson: competitive advantage emerges from the combination of intelligence, context, and the capacity to act.
A decision operating system
From Sinecta's perspective, this discussion connects directly to the design of a Business Operating System (BOS): the set of priorities, owners, metrics, processes, and rhythms through which a company turns strategy into execution.
AI can strengthen that system, but it doesn't replace it. Without clear owners, reliable data, and forums where decisions get made, an organization will only produce recommendations faster.
The cycle
Signals → Options → Scenarios → Challenge → Decision → Execution → Learning.
| Stage | Question it must answer |
|---|---|
| Signals | What changed in the business or the market, and how do we know? |
| Options | What reasonable paths exist, including ones we hadn't considered? |
| Scenarios | What could happen with each alternative, and which assumptions support our estimates? |
| Challenge | Which arguments, data, or risks weaken our favorite option? |
| Decision | Who decides, what gets approved, and which uncertainties are accepted? |
| Execution | Who owns it, with what resources, and against which metrics? |
| Learning | When will we review the decision, and what evidence would make us change course? |
The human component doesn't disappear. It becomes more visible: people define objectives, interpret risks, own the consequences, and determine what deserves the company's attention.
Rhythm matters as much as technology
An organization can generate new analysis every day and still have a committee that reviews the business once a quarter. If analytical capacity outpaces decision-making capacity, the result is information piling up — not competitive advantage.
That's why the design needs a cadence: operating signals in weekly reviews, resource-allocation decisions in monthly forums, and strategic hypotheses reviewed quarterly or when relevant events occur.
The exact frequency depends on the business. The point is that every type of decision has a moment, an owner, and a way to become action.
How to start without turning it into another endless transformation project
There's no need to redesign the whole company. It's better to start with a recurring decision with clear economic impact: setting prices, prioritizing commercial opportunities, deciding growth investments, evaluating acquisitions, or allocating capacity across projects. The pilot can be organized in five moves:
- Pick one decision. Document how it's made today, how long it takes, what information it uses, and where friction appears.
- Connect the context. Identify internal and external sources, access permissions, data quality, and assumptions that need human validation.
- Design the challenge. Require alternatives, scenarios, a counter-argument, and traceable evidence before a recommendation is presented.
- Build the result into operations. Take it to the committee or meeting that already has authority to decide; assign an owner and record what was resolved.
- Measure and learn. Evaluate source quality, the diversity of options considered, analysis time, adoption of recommendations, and subsequent results. Saving hours is useful; better decisions are the goal.
There's a limit worth respecting: AI can also create false confidence. A well-written answer isn't necessarily correct, and systematically delegating judgment can weaken leaders' critical capacity. Other recent Harvard Business Review analyses warn about exactly this tension [3].
The answer isn't to exclude AI, but to design controls: data provenance, independent verification, bias review, documented uncertainty, and human accountability for high-impact decisions.
Strategy should no longer be a document updated once a year
For a long time, formulating strategy meant concentrating knowledge in a periodic process, producing a plan, and hoping execution would follow the expected path.
AI opens the possibility of a more adaptive strategy: exploring continuously, updating models, questioning decisions, and learning from what happens in operations. But that possibility doesn't materialize by buying software licenses.
It requires rethinking how information flows, how alternatives are debated, who can make decisions, and at what rhythm course is corrected.
The future doesn't necessarily belong to the companies that produce the most content with AI. It belongs to those that best turn knowledge into action.
At Sinecta we work at that intersection: strategy, artificial intelligence, and business operating system design. Because a truly AI-native organization isn't the one that uses AI everywhere — it's the one that learns to think, decide, and execute better with it.
[ Author ]
Santiago Rueda
Co-founder of Sinecta. Works with leadership teams on AI adoption, strategy, and business operating systems across South Florida, Mexico, and Colombia.