Of course, plenty of companies will continue to seek out new and innovative ad hoc AI use cases. But the businesses that are truly outperforming their competitors are structuring their AI strategy with a broader portfolio approach. They’re focusing on short-term ROI plays, as well as on longer-term transformations.
Pilots and proofs of concept still have their place, of course, but to unlock true value, scaling is the next frontier in AI — and it’s where most companies stall. Why? Because AI scaling is an evolution that brings new challenges, requires new talent and demands a new mindset around the technology. Here’s how to make sure your business is prepared to make the switch.
What’s behind the move
For many businesses, the experimentation age of AI has been defined by a basic 80-20 approach that looks something like this:
Eighty percent of AI investments are focused on in-house day-to-day elements, such as software development projects and meeting summary tools. These are smaller, more controlled experiments with defined safety guardrails and easily discernible ROI.
The other 20% are broader, more transformational projects, such as large-scale in-house workflow redesigns and customer-facing apps. These bigger leaps generally come with higher budgets, more opportunities for failure and less defined financial outcomes.
Moving toward a more scaled AI approach requires a shift in this ratio that’s inspired by a combination of factors, including the following:
Business trust. As leaders see more success and tangible results from in-house AI experiments, they often start thinking about expanding the bigger transformational swings. They imagine a model for supportable scaling that makes good financial sense in terms of cost and ROI.
Outcome trust. Beyond the financials, leaders start to see reliable, repeatable successes that suggest these results could be amplified on a wider scale without breaking the organization.
Customer demand. It’s not just industry buzz and pressure driving the push toward AI. Consumers looking for more agentic experiences are gravitating to companies that can deliver them — while those that can’t are getting left behind.
Why most AI scaling efforts stall
It might be tempting to look at AI scaling as simply an increase in amplitude. If a smaller-scale experiment worked, after all, why not just replicate it on a bigger scale — more money, more tools, more resources?
AI scaling is about more than expanding; it’s an exercise in redesign.
But moving from AI experimentation to AI scaling requires more than just flipping a switch. Most AI scaling efforts fail not because the models underperform but because companies try to scale experiments instead of redesigning systems.
AI scaling is about more than expanding; it’s an exercise in redesign. It retains the fundamental ideas of what worked, while also planning for what those successes might look like on a much bigger stage. That means building out the organization’s AI capacity in several key areas:
Operating model. Areas like governance, risk management and decision-making need to look different when AI shifts from experimental to scaling. For some companies, that could mean creating new AI-specific roles and shifting team responsibilities to accommodate the step up.
Talent. It’s not just about hiring more engineers; it’s about adding different skill sets as well. A business might have the AI skills to run a small experimental project, but are those skills enough to, say, ensure that AI isn’t injecting bias into a million different transactions? Are those skill sets already in house? If not, where will they come from?
Tools. The AI tools that were good enough for an in-house AI experiment might not be able to handle a bigger and more complex production environment.
Measurables. A faster process might be enough to justify an AI experiment, but scaling requires a deeper look at things like costs, ROI and the business value of retooling workflows around AI.
All these elements come under the umbrella of the one major change that must occur as a business moves from experimentation to scaling: the adoption of an AI-first mindset.
Building an AI scaling culture
Adjusting an operating model, adding tools and talent, reassessing measurables don’t happen overnight. They not only require organizational effort and commitment, but they also demand deliberate planning and preparation relative to the following key factors:
Tech stack. Does the organization’s current configuration allow for scaling? It isn’t particularly difficult to add new tools and capacity, but it’s important to know what is and isn’t in place before moving forward.
Learning culture. AI scaling isn’t a set-it-and-forget-it exercise. The business itself must be built to keep up with the ever-changing world of AI. For example, businesses unaware of the latest innovations in observability might miss some big opportunities.
A nondeterministic mindset. Organizations generally prefer the feeling of safety that comes from being deterministic – that is, “if A, then B.” But that’s not how AI works, and it’s not really how human beings work either. If leaders are still stuck with the expectation of 100% predictability, AI at scale will disappoint.
The difference for humans is that we mitigate our nondeterministic nature through training, review, guidance, scope control, protocols and processes to achieve more deterministic results. AI requires a similarly layered effort that goes beyond just the models.
The businesses leading the way in AI today have moved from experimentation to scaling.
The key is to find a percentage level that’s acceptable for the results the company is trying to achieve — and leaders who can think that way. It might also mean bringing in outside perspectives from partners, peers or clients on whether they see value in how AI is being deployed.
AI is moving forward — are you?
There was a time not so long ago when forgoing AI was acceptable for many organizations. But that time quickly passed. AI experimentation became the new standard. As long as a company was testing out a few AI tools to increase productivity or reduce costs here and there, that was good enough.
But good enough isn’t good enough anymore. The businesses leading the way in AI today have moved from experimentation to scaling.
And those that aren’t prepared for this shift are being left behind.
Rodrigo Domingues is senior vice president of AI engineering at CI&T, a consulting firm that helps businesses navigate business transformation, technology and AI with clarity, speed and impact.
An AI companion sounds dystopian, but it has become a common thread in the wider conversation about the perils of generative AI. What it refers to is essentially a conversational agent built to sustain an ongoing, personal relationship with a user, with the memory and steady persona that keep it consistent from one session to…
A little robotic switch-flipper has become my sidekick in combating cockroaches. Before I got the SwitchBot Bot Rechargeable, I’d tiptoe through the dark every morning, hoping I wouldn’t step on one of those terrible bugs scurrying around as I made my way to the light switch across the room. Now I’m ready for battle before…
Ergonomics is a word that gets thrown around a lot when we talk about peripherals like a keyboard or mouse – but what does this actually mean? The layout and shape, for starters, should adjust to the body’s natural position, and these devices should be engineered to reduce any strains or pains that might arise…
Most engineering leadership teams are grappling with the fact that reliability costs too much, teams are burnt out, and they feel something has to change. Typically, teams resort to optimization: smarter alerting, faster runbooks, better dashboards, and more training. Teams invest heavily, see modest gains, and then end up right back where they started when…
The iPhone Fold’s complexity means production will get off to a slow start.
Steve Dent for Engadget Many people who want Apple’s rumored iPhone “Ultra” may not be able to get it at first, according to a recent industry survey from analyst Ming-Chi Kuo. Kuo believes that demand will be high for Apple’s…
Today’s global economy runs on digital data, yet the United States lacks a national data protection law. The value of technological innovation, from generative AI to predictive analytics, is fundamentally bound to the quality, integrity, and security of its underlying data. In today’s digital economy, data privacy is no longer just a regulatory compliance obligation….
