From Enterprise AI to Agentic Systems: The Evolution of Automation in Business

Enterprises are increasingly adopting agentic systems as they embrace intelligent workflows, according to insights from Databricks. The initial wave of generative AI promised significant business transformation but often delivered only isolated chatbots and stalled pilot programs. Technology leaders struggled to manage high expectations with limited practicality. However, recent data from Databricks indicates a positive shift in the market.

Information from over 20,000 organizations, encompassing 60% of the Fortune 500, suggests a rapid transition towards “agentic” architectures. In these systems, models not only retrieve information but also autonomously plan and execute workflows.

This evolution signifies a substantial reallocation of engineering resources. Between June and October 2025, the utilization of multi-agent workflows on the Databricks platform surged by 327%. This growth signals the graduation of AI to a central element of system architecture.

The ‘Supervisor Agent’ Leading Enterprise Adoption of Agentic AI

A key driver of this growth is the ‘Supervisor Agent’. Instead of relying on a single model for every request, a supervisor acts as an orchestrator, dividing complex queries and delegating tasks to specialized sub-agents or tools.

Since its launch in July 2025, the Supervisor Agent has emerged as the primary agent use case, representing 37% of usage by October. This approach mirrors human organizational structures where a manager delegates tasks to a team. Similarly, a supervisor agent oversees intent detection and compliance checks before directing work to domain-specific tools.

While technology companies are currently at the forefront of this adoption, creating nearly four times more multi-agent systems than any other industry, the utility extends across various sectors. For example, a financial services firm might utilize a multi-agent system to handle document retrieval and regulatory compliance simultaneously, providing a verified client response without human intervention.

Traditional Infrastructure Challenges

As agents progress from answering questions to executing tasks, traditional data infrastructure faces new challenges. Conventional Online Transaction Processing (OLTP) databases, designed for human-paced interactions with predictable transactions and infrequent schema changes, are now being tested by agentic workflows.

AI agents are generating continuous, high-frequency read and write patterns, often creating and dismantling environments programmatically for code testing or scenario runs. The telemetry data reveals the scale of this automation, with AI agents now responsible for 80% of database creation, up from just 0.1% two years ago.

Additionally, 97% of database testing and development environments are currently built by AI agents. This capability enables developers and “vibe coders” to swiftly create ephemeral environments, fostering the development of over 50,000 data and AI apps since the Public Preview of Databricks Apps, experiencing a 250% growth rate over the past six months.

The Rise of Multi-Model Strategies

Enterprise leaders face the persistent risk of vendor lock-in as they expand agentic AI adoption. To counter this, organizations are increasingly adopting multi-model strategies. As of October 2025, 78% of companies were utilizing two or more Large Language Model (LLM) families, including ChatGPT, Claude, Llama, and Gemini.

The sophistication of this approach is on the rise, with the percentage of companies employing three or more model families escalating from 36% to 59% between August and October 2025. This diversity empowers engineering teams to assign simpler tasks to smaller, cost-effective models while reserving advanced models for complex reasoning.

Retail companies are leading the way, with 83% using two or more model families to balance performance and cost. A unified platform capable of integrating various proprietary and open-source models is quickly becoming essential for the modern enterprise AI stack.

Contrary to the batch processing legacy of big data, agentic AI predominantly operates in real-time. The report reveals that 96% of all inference requests are processed instantly.

This real-time processing is especially prominent in sectors where latency directly impacts value. The technology sector, for instance, handles 32 real-time requests for every batch request. In healthcare and life sciences, where applications may involve patient monitoring or clinical decision support, the ratio is 13 to one. IT leaders recognize the necessity for inference serving infrastructure capable of managing traffic spikes without compromising user experience.

The Role of Governance in Accelerating AI Deployments

The relationship between governance and velocity may surprise many executives. While often perceived as a hindrance, robust governance and evaluation frameworks serve as catalysts for production deployment.

Organizations leveraging AI governance tools are able to implement over 12 times more AI projects compared to those without such tools. Similarly, companies using evaluation tools to methodically test model quality achieve nearly six times more production deployments.

Governance establishes essential guardrails, such as data usage guidelines and rate limits, providing stakeholders with the confidence to approve deployment. Without these controls, pilot projects often stall in the proof-of-concept stage due to unspecified safety or compliance risks.

Unlocking Value Through Enterprise Automation with Agentic AI

Although autonomous agents conjure futuristic capabilities, current enterprise value from agentic AI lies in automating routine, mundane, yet critical tasks. The top AI use cases vary across sectors but revolve around addressing specific business challenges:

• Manufacturing and Automotive: 35% of use cases focus on predictive maintenance.
• Health and Life Sciences: 23% of use cases involve medical literature synthesis.
• Retail and Consumer Goods: 14% of use cases are dedicated to market intelligence.

Moreover, 40% of the leading AI use cases cater to practical customer concerns such as customer support, advocacy, and onboarding. These applications drive measurable efficiency gains and build the necessary organizational capabilities for more advanced agentic workflows.

For C-suite executives, the future entails less focus on the “magic” of AI and more on the rigorous engineering practices surrounding it. Dael Williamson, EMEA CTO at Databricks, notes a shift in perspective.

“The conversation in EMEA businesses has transitioned from AI experimentation to operational reality,” Williamson explains. “AI agents are now running critical parts of enterprise infrastructure, and organizations deriving true value are those prioritizing governance and evaluation as foundational elements, not mere afterthoughts.”

Williamson emphasizes that competitive advantage now hinges on how companies build rather than what they purchase.

“Open, interoperable platforms enable organizations to apply AI to their proprietary enterprise data, rather than relying on pre-built AI features that offer short-term productivity gains but lack long-term differentiation,” he adds.

In highly regulated industries, the combination of openness and control is what distinguishes experimental pilots from sustainable competitive advantages.

For more insights: Anthropic has been chosen to develop a government AI assistant pilot

Interested in learning more about AI and big data from industry experts? Explore the AI & Big Data Expo happening in Amsterdam, California, and London. This comprehensive event is part of TechEx and is co-located with leading technology events such as the Cyber Security & Cloud Expo. Click here for more details.

AI News is brought to you by TechForge Media. Explore upcoming enterprise technology events and webinars here.