iTechGuides is reader-supported. When you buy through links on our site, we may earn an affiliate commission. As an Amazon Associate I earn from qualifying purchases. Learn more
The most practical advice in Bruno Guicardi’s September 2026 CIO opinion piece, “Start with the strugglers,” is to begin AI transformation with a team that has a costly business problem and a few people willing to change how they work, rather than with a broad rollout or the highest-performing groups. Guicardi argues that AI transformations stall when companies distribute tools and run short training courses without changing the way work gets done. He presents the approach as practitioner experience, not controlled comparative research, so it is best read as a strong hypothesis with one detailed case behind it.
Guicardi is identified in his CIO contributor profile as a co-founder and president of CI&T, and CIO describes him as a technology transformation specialist. The article appears in CIO’s IT management section.
Why AI rollouts so often fail to deliver
Guicardi opens with a question many leaders will recognize: why do so many AI transformations fail to produce the results they expected? His answer is that the usual program treats AI as a tool deployment. Licenses are distributed, employees complete a course, and success is reported as access, training completion, or adoption. Those measures can all rise while the work itself stays the same.
Free tools Windows power users keep installed
One-click scans. No signup required.
The core of his argument is a single distinction: “AI changes how people work, not simply what tools they use.” If the work process does not change, a new tool sits on top of old habits. The gap between what the company bought and what the team actually does is where most transformations lose momentum.
The sequence Guicardi recommends
The article’s proposed path has six steps. Each one depends on the one before it, which is why the author treats the order as part of the method rather than a menu of options.
- Identify a real business problem that the team already feels, with clear consequences if it is not solved.
- Find a willing risk-taker: people inside the team who are ready to challenge existing work and experiment.
- Provide concentrated, hands-on support so that experienced practitioners work on the live problem alongside team members.
- Work toward a measurable business result rather than an activity count.
- Make the win visible inside the organization.
- Have experienced people help the next team adapt the learning to its own work, instead of copying the first team’s process.
Why start with the strugglers
The phrase “strugglers” is the article’s informal label, not a management category. Guicardi uses it for teams or employees who are under pressure from current conditions. In his telling, these teams are the better starting point for two reasons. They have a compelling need, which gives the change a reason to exist. And a willing group inside them can take on the experiment.
The alternative he questions is starting with a high-performing team that has little incentive to change. Such a team may be productive and content with its current methods, so an AI program asks it to give up something it does not feel it needs. A struggling unit has the opposite problem: the need is obvious, but the people willing to challenge existing work are not guaranteed. Guicardi’s point is that leaders must do both jobs, finding the need and finding the people. Neither is sufficient alone.
Recommended Free Tools
The quoted line that captures this is: “They started with the scufflers, and it made all the difference.” It is the author’s own phrasing, and the choice of a team to start with is the lever the article treats as decisive.
Why a short course is not enough
The article’s second argument is that training alone does not change how work gets done. A course can teach people what a tool does. It does not show them how to restructure a task, when to trust an output, or how to handle the exceptions a real workflow produces.
For that reason Guicardi argues that practitioners should work directly with the team on a real business problem. The support is embedded in the work, not delivered in a classroom before the work begins. This is a costlier model than a rollout, because experienced people spend time inside one team rather than serving many at a distance. The trade-off is that the learning reflects the team’s actual problem, which is what makes it transferable later.
Rank #3
Measuring business outcomes instead of adoption
Guicardi contrasts two kinds of measures. Access, training, adoption, and satisfaction tell you whether people are using the program. Business measures such as revenue, costs, P&L, and market share tell you whether the program is changing results. The article’s position is that the first group is necessary for tracking progress but does not establish that the transformation has worked.
| Question leaders ask | Measures the article treats as insufficient | Measures the article treats as the test |
|---|---|---|
| Is the program reaching people? | Licenses issued, training completed | Not the main test |
| Are people engaging with it? | Adoption counts, employee satisfaction | Not the main test |
| Has the business improved? | Not addressed by these measures | Revenue, costs, P&L, market share |
A practical implication follows: set the business measure before the work begins, so that the team is judged on the problem it was asked to solve rather than on how much it used the tools.
Scaling through visible wins, not a copied playbook
The article’s scaling model is the opposite of a broad rollout. A win in one team is made visible, and experienced people then help the next team adapt the approach to its own problem. The reasoning is that the methods that worked in one unit will not transfer intact, because each team’s work differs.
This is where the approach is most likely to be misread. “Share the win” can easily become “copy the playbook,” which is the pattern Guicardi warns against. Adapting locally means the second team has to redo part of the diagnosis, and that takes time. Organizations that want faster scaling may be tempted to skip that step, and the article does not explain how to keep local adaptation fast.
The bank case: what is and is not established
The article’s main example is a bank that Guicardi does not name. He says the bank assigned 100 AI experts to work alongside 100 client employees on an investment-team effort, and that the team reversed three consecutive years of market-share losses within 12 months. He also says the bank moved experienced investment-team employees into other teams, and that the CEO publicly recognized the result.
These are the author’s own reported figures and claims, presented in an opinion article from 2026. They have not been independently audited, and the bank’s identity is not given, so the staffing levels, the timeline, and the attribution of the market-share recovery to the program cannot be checked from the article alone. They show what the approach looked like in one organization. They do not show that the same approach produces the same result elsewhere.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.How to test the approach before committing to it
Guicardi’s argument is most useful as a set of checks a leader can apply to any proposed AI program:
- Can you name a specific business problem in one team, with a measurable cost of leaving it unsolved?
- Do you know who in that team is willing to change the existing work, and have you asked them?
- Will experienced practitioners work on the live problem, rather than only delivering training?
- Is the success measure a business result, not access, adoption, or satisfaction?
- Is there a plan for how the next team will adapt what was learned, rather than copy it?
If the answer to several of these is no, the program is probably following the rollout pattern the article criticizes, whatever its stated goals.
The limits of the argument
The article is an opinion piece by a practitioner who has a commercial stake in the transformation approach his firm works in. It is persuasive on the diagnosis: tools and training do not by themselves change how work is done. It is thinner on evidence for the prescription. It does not report results from multiple organizations, does not describe how to measure the counterfactual, and does not explain how to find willing people in teams that may be defensive. Those are the questions a reader should answer for their own organization before adopting the sequence.
Do these 3 things before closing this tab:
1Repair Windows errors before they cause bigger problems2Fix the driver behind crashes, sound loss and screen glitches3Clear out junk files and repair common Windows errorsThe most defensible reading is that starting with a real problem and a willing team, supported on live work and measured by business results, is a sound way to run an AI program. Whether the struggling-team starting point is better than other choices is a claim the article supports with one case and its author’s judgment.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
