Agentic AI is 'Table Stakes', Survey Finds
A recent developer survey indicates that agentic AI is rapidly becoming a core expectation in software development. The survey found that 85% of developers expect AI agents to be "table stakes" within three years. Underscoring the importance of infrastructure, 94% of respondents said they would switch cloud vendors for better support for agentic AI.
- Agentic AI differs from generative AI by focusing on autonomous action to achieve a goal; it uses large language models as a "brain" to orchestrate and execute tasks with minimal human input, rather than just creating content. - The survey from Nylas also found that 67% of developers are already building or shipping agentic workflows, indicating a rapid shift from experimentation to production. - Major cloud providers are heavily invested in this "infrastructure race," with Google Cloud launching "Agentspace" for building custom agents and AWS releasing "Strands Agents," an open-source SDK to simplify agent creation. - A separate Salesforce study of over 2,000 software leaders found that 92% of developers believe agentic AI will help them advance in their careers, with most eager to use agents for debugging, error resolution, and generating test cases. - However, the 2025 Stack Overflow Developer Survey indicates some hesitation, with 38% of developers stating they have no plans to use AI agents, and overall trust in the accuracy of AI tools dropping from previous years. - The San Francisco Bay Area remains the center of AI investment, attracting over $200 billion in AI funding between 2020 and early 2026, with corporate investors like Google, Microsoft, and Amazon now accounting for 40% of the total. - Startups are already shipping agentic products, including Cognition AI's coding agent "Devin" and Qodo, an enterprise platform that uses a multi-agent system for AI-driven code reviews. - One real-world example of agentic AI adoption is the productivity platform ClickUp, whose CEO reported having over 3,258 AI agents working alongside 1,300 human employees to manage workflows.
