Tech executives are rapidly adopting agentic artificial intelligence, with a recent survey revealing that nearly half of the respondents have already begun implementing such systems in their organizations. The Technology Pulse Poll conducted by Ernst & Young found that about 50% of tech leaders anticipate their internal artificial intelligence operations will be fully autonomous within the next two years. The survey, which included over 500 technology leaders, indicated that 58% believe their organizations are ahead in AI adoption. However, Ernst & Young’s Americas Technology Sector Growth Leader, Ken Englund, cautioned that these perceptions may not align with reality, suggesting that companies often overestimate their progress. Despite concerns about potential job losses due to AI, 84% of tech leaders plan to hire more staff in the next six months as they integrate new AI tools, while over half are focusing on upskilling their current workforce. Notably, data privacy and security emerged as the top concerns for 49% of respondents, reflecting a growing awareness of the risks associated with AI deployment.
Executives are heavily investing in artificial intelligence, yet a recent survey by IBM reveals that many are dissatisfied with the returns on these investments. Only 25 percent of the 2,000 CEOs surveyed reported that their AI initiatives met expectations, and just 16 percent have scaled AI across their entire organizations. Despite these challenges, a significant 85 percent of CEOs expect positive returns on their AI investments by 2027. However, communication around AI strategies remains weak, with only 15 percent of U.S. employees feeling their organizations have clearly articulated an AI strategy. As companies continue to invest tens of billions into AI, the path to profitability remains uncertain.
And I wanted to share this example. AT&T has successfully developed a cost-effective open-source artificial intelligence system to categorize its annual 40 million customer service calls, significantly reducing reliance on expensive models like ChatGPT. The new system, which combines several smaller open-source models, now processes call summaries in under five hours, compared to the 15 hours required by the previous method. According to Hien Lam, a senior data scientist at AT&T, this innovative approach not only cuts costs by 65% but also maintains a high accuracy rate of 91%. By using a combination of models tailored to specific tasks, AT&T aims to enhance customer service efficiency while safeguarding user data.
Why do we care?
Nearly half of tech executives say they’re implementing agentic AI, and 50% expect full autonomy in two years. But contrast that with IBM’s findings: only 16% have achieved any significant scale. There’s a disconnect between ambition and execution, and that’s where IT service providers (especially those acting as Strategic Outcome Providers) can step in—to bridge the gap between vision and real, working systems.
The AT&T example is the exception, not the rule. They customized smaller open-source models, targeted a defined problem (categorizing 40M service calls), and achieved measurable ROI: 65% cost savings and faster processing. This kind of focused, scoped, and data-governed AI deployment is what most organizations lack—and what IT services firms can help deliver.