Globant (GLOB) AI Agents Deliver Software 30% More Productively

Globant S.A. (GLOB) said on its August 13, 2026 earnings call that AI agents supervised by engineers deliver software with more than 30% higher productivity than the typical "engineer plus AI" approach [1]. The company did not disclose how that figure was measured, and it did not give the effect on gross margin.

What the AI application does

Globant S.A. (GLOB) provides software development and digital transformation services to enterprise clients. It makes money by delivering engineering teams to customers, and engineering labor is its largest cost. This AI application sits inside that core business. Globant places a set of AI agents, supervised in real time by its own experts and by the client, into client software projects. The agents handle work such as mapping requirements, designing architecture, writing code, fixing defects, testing and modernizing legacy systems.

Globant has sold this capability first as the CODA agent suite and then as AI Pods, dedicated delivery units that clients buy on subscription. In the latest disclosure, it sits within the company's Glob.AI services framework. Before the agents start work, the platform automatically assembles context, maps the architecture, prepares data and sets quality gates. Pricing is based on output or usage rather than hours or seats [1].

From internal tool to subscription service in two years

Over two years the application moved from an internal development tool to a core service sold on subscription.

• August 15, 2024. Globant first described AI agents covering product definition, back-end prototyping, design, testing and code fixes. Management also cautioned that code generation tools improved developer productivity by less than 15%, and single tasks such as unit testing by about 40%. Gains on individual tasks do not necessarily add up at the team level, and the company said it cares about overall team performance [2].
• May 15, 2025. The agents were packaged as the CODA suite and sold to clients as subscriptions through AI Pods, with work metered in tokens. Management said such a subscription could be much cheaper than a team of five engineers, but said the related deals were still small and did not disclose figures [3].
• February 26, 2026. The company said more than 60 AI Pods were running with clients worldwide, with 24 new subscriptions signed in the prior quarter. It cited a FIFA case in which engineering delivery costs fell 20% without affecting velocity or quality [4].
• August 13, 2026. AI Pods were folded into Glob.AI, and the company for the first time gave an overall productivity figure against an "engineer plus AI" baseline [1].

How to read the 30% productivity figure

The latest productivity figure now has a comparison baseline, but its scope is still loose. The company said productivity is more than 30% higher than the typical "engineer plus AI" approach [1]. Because the baseline already uses AI assistance tools, the number measures the extra productivity of the agent workflow over ordinary AI-assisted development. Globant did not say which tasks, how many projects or what period it covers, so it cannot be read as Globant's company-wide labor productivity.

Management also gave an example. A client's complex ERP and API project needed architecture design; the team met with the client two or three times and finished in 48 hours, work that management said would take at least two to three weeks the traditional way [1]. That is one task for one client, and the two-to-three-week figure is a management estimate.

The earlier numbers need the same care. The more than 60 AI Pods include operations Pods. The 20% cost reduction at FIFA happened on the client side, and neither the cost base nor the period was disclosed [4].

Will the gains reach Globant's gross margin?

Whether these gains become Globant's own profit depends on gross margin. If the agents cut the hours needed for the same deliverable while output- or usage-based pricing does not fall in step with hours, the labor savings could stay with the company and lift gross margin. If the gains mainly pass to clients as cheaper subscriptions and client cost savings, the more likely result is lower revenue per project, and gross margin may not improve.

So far, most of the disclosed results sit on the client side. The FIFA figure is a reduction in the client's delivery costs [4], and the five-engineer comparison says the client can pay less [3]. Globant has not disclosed revenue, pricing, delivery headcount or gross margin contribution for AI Pods or Glob.AI. Gross margin is also affected by pricing, staff utilization, compensation and exchange rates, so any change in it cannot yet be attributed to this application.

What is confirmed and what is not

What can be confirmed is that AI agents are now part of Globant's main delivery business and have started to change how it charges clients. What has not been shown is whether that change has altered the company's profit structure. The most useful next piece of evidence would be disclosure of revenue and gross margin for output-based pricing, compared with traditional time-based projects.

Application assessment

• AI Software Engineering Delivery | Business position: core business | Stage: limited production | Coverage: multiple businesses or regions | Value type: cost reduction

Sources

[1] Drillr · Globant S.A. (GLOB) · 2026-08-13 · Earnings call

"We have seen plus 30% more productivity than with a typical engineer plus AI approach."

[2] Drillr · Globant S.A. (GLOB) · 2024-08-15 · Earnings call

"But the problem with junior developers is that it brings very good quality code compared to what they would have done without those tools. But if you say as a unit, it is like a car spare part, it looks super good, but it doesn't fit the car. So it's kind of tricky. What's important to us, it's overall team performance. It's not individual. And we're measuring like things that when you put all of that together, they don't necessarily add up."

[3] Drillr · Globant S.A. (GLOB) · 2025-05-15 · Earnings call

"So whenever you had a team of five engineers, we call it that way, now you may have a subscription that is much cheaper than those five engineers. But on the back, there's the set of agents producing the software supervised by a human that allow that the quality and the output doesn't have hallucination, and it's the same Globant quality as always."

[4] Drillr · Globant S.A. (GLOB) · 2026-02-26 · Earnings call

"Through the deployment of AI pods, we were able to move beyond traditional consulting services and achieve a major financial milestone for the organization, reducing costs by 20% without compromising the velocity or quality of our engineering output."