Leaders and specialists in corporate technology gathered at the Smarter.tech '26 event to discuss a complex question: what does it truly mean for organizations with existing complex infrastructure and overloaded teams to work more efficiently when they are now responsibly implementing AI?
The general conclusion of the event was an emphasis on context. Technology only becomes useful when it aligns with people's working environment, respects existing organizational controls, and provides teams with a practical way to operate it.
Muggi van Staden, CEO of Obsidian Systems, set the tone with his opening speech, urging participants to actively engage in the program so that everyone could gain a clearer picture of what smarter technology looks like in practice.
Van Staden noted: 'Clients are not asking for more isolated technologies. They want to know how it will fit into their environment and how it will be managed. They also need assurance that their teams can reliably manage and maintain it.'
He added that Smarter.tech '26 provided an opportunity to include these questions in one conversation with the tools and platforms shaping the market.
The Atlassian plenary session demonstrated why context has become a central element in enterprise AI. Their direction for Rovo and Teamwork Graph is based on linking organizational knowledge with the work already being done in Jira, Confluence, and other related systems, giving AI agents access to relevant information while maintaining human involvement in the process of assignment, verification, and completion of work.
Chris Mulder from Red Hat presented another aspect, describing the challenges faced by organizations that must maintain legacy workloads alongside containers and new AI platforms, often without increasing teams or budgets. His solution involved a consistent operational model built on open platforms, which ensures greater control over where workloads run and how data is used.
This idea continued in practical demonstrations conducted by Obsidian teams. One session showed how Red Hat Satellite and Ansible can replace disparate Linux server management with a managed operational model covering deployment, troubleshooting, and reporting. The importance lay not so much in the tools themselves, but in the expertise required for their integration and the discipline needed for their maintenance.
This is a familiar problem for South African organizations modernizing relatively old systems, facing skill limitations, growing security threats, and unpredictable technological costs. Simply adding a platform without first defining how it will be managed and maintained only increases complexity.
Van Staden concluded that 'smarter technology should reduce the burden on people, not give them another system to manage.' He stressed that this requires technical depth and strong partnerships underpinned by an understanding of the client's actual operational activities.
Smarter.tech '26 asserted that progress would be achieved not by chasing every new possibility, but by making better decisions about how technologies connect, who is responsible for them, and how the environment can be improved over time.
Employees of Meta Platforms filed a lawsuit in the United States, alleging that the company's artificial intelligence systems were responsible for selecting employees to be laid off. This litigation raises important questions about the challenges of proving the role of these technologies in employment decisions.
The dispute involves 26 employees who claim to have suffered harm due to criteria related to disabilities, family leave, or medical absences. The lawsuit, which is proceeding in the US, seeks a temporary suspension of the layoffs while the arguments presented by the parties are examined.
A central obstacle for the workers is proving how the AI tools were implemented internally by the corporation. Without access to the operating systems and the parameters used, the employees find it very difficult to gather robust evidence to support their claims.
This lawsuit against Meta is seen as one of the first legal actions to directly challenge the use of artificial intelligence in the workforce reduction process. The case also helps clarify why public disputes over technology-assisted labor decisions are still uncommon.
Experts cited in the case indicate that workers often have little knowledge about how companies use automated systems. Additionally, many professionals are subject to arbitration agreements, which divert certain conflicts from traditional justice to private procedures.
This combination of restricted access to information and procedural limitations makes it difficult to file class-action lawsuits. In these scenarios, employees risk losing the opportunity to take the case to a jury or seek compensation in public forums.
Lawyer Christine Webber, associated with civil rights and labor relations at the firm Cohen Milstein Sellers & Toll, emphasized that even if flaws in a system are identified, the information might remain restricted to the individual case. She told Reuters that even if there was proof that a specific system would repeatedly generate discriminatory results, there would be no way to disseminate this information to other employees. It is important to note that Webber does not represent the employees involved in the lawsuit against Meta.
Another example of artificial intelligence application in the workplace involves Workday, a human resources management software provider. This company was accused of operating a tool that allegedly eliminated job candidates based on attributes such as race, age, and disability. Workday refutes such accusations, and in this case, the issue of arbitration does not apply because Workday does not maintain such agreements with the clients' candidates.
In the lawsuit against Meta, the employees are seeking a court order to temporarily suspend the dismissals. Although workers' contracts provide an exception for urgent requests to the Court, such a measure is more commonly used in disputes related to trade secrets, not personnel cuts.
Initially, Judge William Orrick denied the request to block Meta's layoffs. His decision was based on the lack of sufficient evidence presented by the workers to contradict the company's version, according to which the choices were made by people, not by automated systems. However, the judge kept open the possibility of reevaluating his decision if evidence of improper use of artificial intelligence emerged. A hearing was scheduled for August 24, and both parties have the right to appeal.
The employees argue that Meta employed various internal AI tools during employee evaluations, including systems capable of monitoring productivity and the use of technological resources. They allege that these analyses harmed individuals who were absent for medical reasons or family responsibilities.
Among the tools mentioned in the petition is an internal assistant named 'Metamate,' described by the plaintiffs as an artificial intelligence-based support for employees. The petition also cites a productivity scoring system that would analyze data such as keyboard logs, screen content, emails, and browsing history.
Meta vehemently contests this interpretation. In documents filed with the court and in statements following the start of the process, the company stated that the decisions regarding the nearly 8,000 layoffs announced during that period were made by human beings. The company also denied that the use of artificial intelligence served as a criterion for selecting who would be dismissed or for evaluating performance.
The workers' lawyers acknowledge that the biggest difficulty lies in accessing the company's internal data. In a joint statement, they requested that both current and former Meta employees provide information on how the artificial intelligence tools were used in determining the cuts.
