AI agents move beyond pilot phase to become a staple in businesses
2–3 minutes
From experimentation to large-scale implementation
Over the next twelve months, discussions about artificial intelligence in corporations are expected to shift from testing to issues of scale, control, and financial returns. A PwC study of U.S. companies found that 79% already use AI agents, and 88% plan to increase investment in this technology.
Integration into daily operations
Where agents first appear
A Cloudera survey of approximately 1,500 technology leaders across 14 countries found that 57% have implemented agents in the past two years, with 21% doing so just last year. This indicates that the technology has moved beyond being a standalone project and has become integrated into the daily operations of many organizations. The first use cases are emerging in areas such as customer service, finance, and logistics. A common example is invoice reconciliation, where the agent cross-references documents, identifies discrepancies, and forwards only questionable cases for human review.
The limits of trust
However, adoption is not uniform. According to PwC, only 20% of executives trust agents for financial transactions, while 38% trust them for data analysis. Higher-risk activities typically require mandatory human oversight.
Governance and compliance take on greater importance
As usage expands, questions arise about who is responsible for decisions made by machines. PwC found that only 45% of companies reevaluate their operational models, and 42% redesign processes around agents. In this context, governance means establishing clear rules: what data AI can access, who approves each step, and how decisions are recorded. Brazil’s General Data Protection Law (LGPD) makes this control a requirement for companies. An IBM survey found that 82% of senior executives believe that silos between departments hinder value creation.
New skills become essential
As agents take over repetitive tasks, the demand for skills that machines cannot replace is growing. These include the ability to formulate effective instructions, evaluate response quality, identify errors, and understand the tool’s limitations. For example, a human resources analyst using AI to filter resumes must check for potential biases before presenting a list to management. This type of responsibility is likely to become more common across various job roles.
Outlook through 2027
The four trends described—agents in operation, AI integrated into systems, structured governance, and team skill development—are expected to progress together. Studies indicate that technology adoption is advancing faster than process reorganization and employee training. For professionals, the most useful indicators to monitor are the budget allocated to technology, changes in work routines, and the availability of training within companies.
