Globant’s new pitch is easy to understand: buy a result rather than a block of staff hours. In August 2026, the company introduced Glob.AI, a service it says lets enterprise customers deploy “AI Pods” made of agents supervised by human specialists, with charges tied to output or consumption. That is a change in how software work is packaged and sold. It also raises a basic question that a pricing page cannot answer on its own: who decides whether the output is good?
Co-founder and CEO Martín Migoya framed the launch as a shift away from hours and seats. The company’s August announcement describes the pods as specialized by task and industry. A later leadership release named Sarab Narang CEO of Glob.AI, reporting to Migoya. Those are company claims and appointments, not independent proof that every project achieves the speed or quality its marketing anticipates.
For a client, output-based pricing sounds cleaner than counting workers at keyboards. Yet software is full of results that look complete until they meet a real user, a security review or a system they must integrate with. If human experts supervise the agents, their judgment remains part of the product. It must be specified, funded and accountable. Otherwise, the apparent simplicity of a finished “output” can conceal review work that is every bit as important as the automated step.
Globant’s own second-quarter figures complicate any story of effortless growth. Its August 13 report put revenue at $614.4 million for the quarter and headcount at 27,411 at June 30, down from 30,084 a year earlier. The filing describes workforce resizing and office reductions in business optimization programs. It does not establish that AI caused each lost position, and it would be false to turn the two numbers into that claim. But they make the labor dimension of a new services model impossible to leave out.
The company was founded in Latin America and now sells globally; its report says just over half of second-quarter revenue came from North America and about one-fifth from Latin America. That split matters for a regional technology workforce. The commercial promise may be negotiated with enterprise buyers far from the teams doing the work. If clients pay for output, the people supervising it need clear authority to reject weak or unsafe results, not merely a mandate to move faster.
Glob.AI may change the unit on an invoice. It cannot eliminate the human choices inside that unit: what to build, what to check, who bears a failure and who gets credit for a success. A serious account of this Latino-founded technology company should follow those choices as closely as it follows the new product name.