Agent confidence on the technical frontier
A new report reveals high confidence among technology experts in using agentic AI for various tasks, particularly in data workflows, driven by the promise of automation and ROI.
Intelligence analysis by Gemini 2.5 Flash
Enterprise investment in AI is surging, with executives looking to agentic AI to deliver measurable financial outcomes. A survey of 300 global technology experts indicates strong confidence in agents for AI, data, and cloud tasks, though readiness drops when agents lack sufficient business context for complex decision-making.
Imagine you have a super-smart robot helper for your computer tasks. People who build and use these robots are becoming really confident that they can do a great job, especially with tasks that involve lots of numbers and data, like checking if all your toys are counted correctly. But for the robot to do its best work, it needs to understand *why* it's doing something, not just *what* to do. So, giving it enough background information, like telling it the rules of your game, is super important, and grown-ups still need to watch over it to make sure it's doing things safely and correctly.
Analysis
Surging Confidence in Agentic AI
The landscape of enterprise AI is undergoing a significant transformation, with agentic AI emerging as a key driver for achieving measurable financial outcomes. A recent report, based on a survey of 300 global technology experts, highlights a burgeoning confidence in the capabilities of AI agents across a wide array of technical tasks. This confidence is particularly high for routine and measurable tasks, such as generating reports or boilerplate code, where agents can streamline processes and reduce repetitive work. The findings suggest that as organizations seek to align their AI projects with strategic business objectives, agentic systems are increasingly viewed as essential tools for improving performance and operational efficiency.
Beyond simple automation, the research indicates a growing trust in agents for tasks requiring more complex judgment and multi-step workflows. This expansion of confidence into more intricate domains underscores the evolving sophistication of agentic AI and its potential to tackle challenges that previously demanded significant human intervention. The report positions 2026 as an "inflection year" for organizations to strategically integrate AI, with agents playing a pivotal role in this shift by managing and coordinating entire workflows, fostering a collaborative environment between humans and AI.
Data Workflows as a Breakthrough Domain
One of the most promising areas for agentic AI, according to the survey, is within data workflows. Technology teams exhibit the highest level of trust in agents when dealing with structured data, which provides a reliable foundation for automated decisions. Specific applications where agents are proving invaluable include data quality monitoring, visualization anomaly detection, real-time data stream monitoring, and data profiling. This domain benefits immensely from the ability of agents to process and analyze vast amounts of information with precision and speed, identifying patterns and anomalies that might elude human observation.
The success in data workflows is attributed to the ability of domain experts to provide crucial context at the point of data generation. This contextual input allows agents to act with greater accuracy and deliver trusted outcomes, making data-centric tasks a breakthrough use case for agentic AI. The structured nature of data, combined with expert guidance, creates an optimal environment for agents to perform reliably and securely, thereby enhancing the overall integrity and utility of enterprise data assets.
The Critical Role of Context and Oversight
Despite the surging confidence, the report identifies a significant challenge: the lack of sufficient business context supplied to agentic systems. As tasks become more complex, agents require greater reasoning capability and a deeper understanding of the business environment to perform effectively. The current stage of development for context-generation capabilities, especially when integrating disparate enterprise data, is still early, posing a hurdle to widespread deployment at the speed and quality demanded by developers and executives.
Human oversight remains a key factor for the successful deployment of agentic AI. Experts emphasize that teams cannot delegate critical work to agents without absolute confidence in their capability, safety, reliability, and security. Integrating agents within existing operational boundaries, identity systems, and governance models, as highlighted by Microsoft Azure's Jeremy Winter, is crucial for building trust. As experience with agents deepens and business environments mature, the expectation is that agent confidence will accelerate, provided these contextual and oversight challenges are adequately addressed.
Key points
- Enterprise investment in AI agents is booming, driven by the need for measurable financial outcomes and reduced IT costs.
- Technology experts show high confidence in agentic AI for various tasks, with data workflows identified as a breakthrough domain.
- A primary challenge for agent readiness in complex tasks is the lack of adequate business context provided to the systems.
- Human oversight and integration into existing operational boundaries and governance models are crucial for successful and trusted agent deployment.
- Confidence in agents is expected to accelerate as tech teams gain more experience and business environments mature.
The increasing confidence in agentic AI suggests a future where complex workflows are largely automated, significantly reducing IT infrastructure costs and freeing human tech teams to focus on strategic innovation. This could lead to improved operational performance and a substantial return on AI investments for businesses.
If agentic AI systems continue to lack sufficient business context for complex tasks, or if context-generation capabilities remain underdeveloped, there's a risk of unreliable or unsafe automated decisions. This could hinder widespread adoption, erode trust, and prevent organizations from realizing the full ROI potential of their AI investments.
