The global use of AI agents grew by 327% in just four months, positioning them as a key emerging capability in AI adoption, according to Databricks. This scenario shows significant progress in the path of agentic AI, scaling from theoretical models to integration backed by measurable business results.

The transition from traditional large language models (LLMs) to multi-agent systems responds to the need to automate complete and specialized workflows, rather than isolated tasks. According to the 2026 State of AI Agents report, this exponential growth is attributed to multi-agent architectures shaping a “new business operating model.”

An example of these scenarios lies with organizations integrating “Supervisor Agents”, a Databricks solution that orchestrates various agents and tools to execute highly complex processes, making systems not only able to reason but also to plan and execute actions independently based on the specific context of the company. After its launch in July 2025, the use of the Supervisor Agent became the main use case, accounting for 37% of total activity on the Databricks platform by October 2025.

Since the massive emergence of generative AI (GenAI) three years ago, industries have moved from an exploratory approach to one of critical operational results. Currently, 66% of organizations already use AI-powered tools, although the implementation of autonomous agents remains at 19%, indicating significant room for scalability in the coming years.

The current relevance of AI agents lies in their ability to bridge the gap between technical potential and business value. This phenomenon is occurring at a time of “once-in-a-decade” infrastructural change in the data layer, according to the Databricks report. The report also agrees that leading companies have moved away from seeking a “single model” to adopting multi-model flexibility strategies, with 78% of companies using two or more model families to optimize performance and costs depending on the task.

Data Layer Automation and “Vibe Coding”

One of the most disruptive findings in the report is the autonomy that agents have achieved in infrastructure management. Telemetry data indicates that:

80% of current databases are created by AI agents, a dramatic increase from 0.1% in 2023.

97% of test and development environments (database branches) are built by agents, reducing provisioning times from hours to seconds.

This advance is driven by the emergence of “vibe coding,” where users describe requirements in natural language for AI to generate the corresponding code. Gartner estimates that by 2028, 40% of new production software will be created using these techniques. To support this load, new-generation operational databases such as Lakebase have emerged, designed to handle the high read and write frequency generated by autonomous agents.

Evaluation and Distribution by Industry 

The report also identifies a direct correlation between the use of control tools and project success. The market has responded with a sevenfold increase in AI governance investment over a nine-month period. 

The efficiency metrics are compelling. Companies that apply unified governance protocols manage to put 12 times more AI projects into production than those that do not. Likewise, the use of systematic evaluation tools (customized benchmarks) allows almost six times more projects to be brought into the production environment, ensuring the accuracy and security of responses.

The technology sector leads adoption, building almost four times more multi-agent systems than any other industry. However, the application of AI has diversified:

Retail is the sector most prone to multi-model use, with 83% adoption of two or more LLM families.

Customer experience represents 40% of global use cases, covering technical support, onboarding, and personalized marketing.

Latin America shows a pragmatic approach with loan origination as the main use case (10%). In addition, the region processes 77% of its inference requests in real time, underscoring the importance of low latency in emerging markets.

The report concludes that the future of enterprise AI is moving away from selecting isolated models to focus on integrating business context and autonomous execution tools. Multi-agent systems are expected to evolve toward continuous learning models, where real-time assessments allow agents to adapt to changes in the environment without constant human intervention.