A recent survey revealed that 88% of US companies plan to increase their AI-related budgets in the next 12 months, driven by the adoption of agentic AI. According to a PwC survey of 300 senior executives, 79% of these firms have integrated AI agents into their operations, with 66% reporting measurable productivity gains.
Despite this widespread adoption, only 35% of companies have broadly implemented AI agents, and just 17% have fully integrated them into almost all workflows.
The survey also highlights that 75% of executives believe AI agents will transform workplaces more than the internet did. Furthermore, 71% anticipate that artificial general intelligence will become a reality within two years. However, 46% of respondents express concern that their companies may fall behind competitors in adopting AI agents.
Organisational challenges pose a significant barrier, with 34% of executives citing cybersecurity and cost as top challenges, while only 19% mention the ability to connect AI agents across applications and workflows. Trust in AI agents remains an issue, with only 20% of respondents trusting AI for financial transactions, despite 38% trusting it for data analysis.
While 18% of companies are not using AI agents, citing a lack of clear use cases, those that have adopted them are seeing tangible benefits. More than half plan to expand AI agent use in customer service, sales and marketing, and IT and cybersecurity within the next six months. However, fewer than half are rethinking operating models or redesigning processes around AI agents.
To stay competitive, businesses need to move beyond pilot projects and fully integrate AI agents into their operations. This involves reimagining business strategies, focusing on workforce transformation, and orchestrating multiple AI agents across complex processes. The survey revealed that building trust and a responsible AI foundation is crucial as AI agents take on more autonomous roles.
Microsoft CTO outlines future of AI agent cooperation
In another development, Microsoft is advancing towards a future where AI agents from different companies can interact seamlessly. At a media briefing held at Microsoft’s headquarters in Redmond, Washington, chief technology officer Kevin Scott discussed the company’s efforts to encourage the adoption of industry standards that would allow AI agents from various creators to work together. The AI systems are designed to independently execute specific tasks, such as software bug fixes.
Microsoft is endorsing the Model Context Protocol (MCP), an open-source standard introduced by Anthropic, which has Google’s backing. Scott likened the potential of MCP to the transformative effect of hypertext protocols in the 1990s, suggesting it could create an “agentic web” for AI collaboration.
Additionally, Microsoft is working on enhancing the memory capabilities of AI agents. Scott pointed out that current AI systems lack the ability to remember past interactions, which results in a transactional approach. Improving memory functions demands substantial computing power and incurs high costs. To address this, Microsoft is exploring structured retrieval augmentation, a technique where AI agents summarise each user interaction to develop a reference guide for future tasks.
Read more: Nearly half of tech firms now implementing agentic AI, survey finds
