The most important AI capability going mainstream in the next 12 months
The most important capability will be agentic AI moving from experimentation into everyday business workflows. The first wave of AI was largely about generating content, summarizing information or answering questions. The next wave is about AI that can actually take action inside a workflow.
A basic chatbot can answer a question, but agentic AI can understand where someone is in a process, determine the next best step, trigger a follow-up, coordinate across systems and know when to escalate. That’s where AI starts to become operationally meaningful.
But the real opportunity is not just adding an agentic layer and calling it innovation. Everyone will have access to powerful models in the next year, but what are you doing with them that is specific, valuable and hard to replicate? Companies will start moving past generic AI use cases and toward agentic AI that’s tied to real workflows, real data and real business outcomes.
How AI changes organizational decision-making by 2027
That will ultimately change how organizations make decisions. They’ll look less at isolated tools and more at ecosystems and how their software, data, partnerships, workflows and customer relationships fit together.
I also think AI will push companies to rethink fragmentation. In the past, owning a separate piece of the stack could be a source of defensibility. But in an AI era, where basic features are easier to replicate, fragmented tools may not be enough. Companies will need to partner more strategically, integrate more deeply and build connected ecosystems that create more value for customers and are harder to reproduce from the outside.
The most underestimated AI risk / bet leaders will get wrong
Assuming AI is correct just because it sounds confident. AI can produce a polished answer even when it lacks the full context. That’s a major problem because many business questions are not universal. The right answer often depends on the customer, the policy, the market, the data source or the specific situation.
That’s especially true in high-stakes industries like housing, finance, healthcare or legal services. In leasing, for example, an answer might depend on the property, the owner, the unit, the applicant’s situation or a local policy. AI can’t just improvise in those moments. It needs approved knowledge bases, clear guardrails and a defined path to a human when the question becomes sensitive or nuanced.
AI is very strong when the task is deterministic, meaning the answer should be consistent and the workflow is repeatable. But not everything in business is deterministic. Some decisions require judgment, context and human agency. The companies that understand that distinction will be in a much better position than the companies trying to automate everything at once.
Advice for a company starting its AI journey today
Start by asking, “Where can AI create real value, and where do we still need human judgment?”
The best early use cases are repeatable workflows with clear rules, trusted data and measurable outcomes. AI can be extremely powerful in those areas because it can reduce manual work, speed up response times and help teams operate more efficiently. But if the workflow is ambiguous, sensitive or heavily dependent on context, companies need to be much more careful.
I would also tell companies not to mistake an AI layer for a defensible strategy. Simply putting AI on top of an existing product is not enough. The underlying tools are becoming more widely available, which means the bar is rising for everyone. The advantage will come from domain expertise, proprietary context, deeper integrations, strong partnerships and workflows that are difficult to replicate.
AI is going to force companies to think more deeply about their core value. What do you do better than anyone else? What do you know that a generic model doesn’t know? What part of your ecosystem would take years for someone else to rebuild?
The AI trend you are most excited about
I am most excited about the shift toward domain-specific AI. Generic AI is impressive, but becomes game-changing when it understands a specific industry, a specific workflow and the actual problems customers are trying to solve.
I also think this trend will force excellence across the market. It will not only pressure traditional software companies, but also pressure AI companies that are mostly repackaging commoditized model capabilities. As the underlying tools become available to everyone, every company will have to answer the same question: what are you really doing that is specialized, defensible and valuable?
That’s what makes this moment so interesting. AI is democratizing access to capability, but it’s also raising the standard.
