May 2026 Summaries

The Background Agents summit, featuring speakers from companies like Stripe, Uber, and Cloudflare, highlighted the growing significance of background agents in software engineering. The summit explored how these agents, emerging as a real category, address the gap between individual developer speed and organizational velocity by focusing on coordination and infrastructure rather than just coding speed. Companies independently developed similar architectures using five key primitives: sandboxed environments, context connectivity, triggers, fleet orchestration, and governance, leading to a convergence of ideas. Notably, organizations like Uber and Monzo demonstrated that starting with foundational improvements, rather than chasing novel applications, can accelerate AI adoption, while the summit revealed unexpected insights such as the broader impact of background agents across industries like genomics and their potential to democratize codebase participation. Cloudflare's example of transitioning 93% of its engineering organization to delegated engineering using these primitives underscores the transformative potential of background agents, making them a crucial infrastructure decision for engineering leaders.

The AI-SDLC Framework is a strategic tool designed to guide engineering leaders through the transition of integrating AI into software development, moving beyond basic tools like Copilot to a comprehensive strategy. It outlines three stages of AI integration: "In the loop," where AI assists developers; "On the loop," where developers orchestrate AI agents; and "Autonomous loops," where AI agents operate independently, with human oversight. The framework emphasizes the importance of understanding and addressing bottlenecks that arise at each stage, such as review processes, organizational capabilities, and security concerns. It encourages organizations to honestly assess their current stage, identify future constraints, and incrementally enhance their AI integration while ensuring security measures keep pace with increased autonomy. The framework facilitates meaningful conversations among leaders, helping them shift focus from tool selection to strategic planning, ultimately aiming for a seamless AI-driven transformation in engineering processes.

AI adoption in large enterprises faces challenges in moving from pilot projects to widespread production, often resulting in unstructured data and high costs without clear returns on investment. The traditional method of deploying forward-deployed engineers (FDEs) can falter as these engineers often leave without imparting sustainable knowledge. Instead, a balanced approach involving internal champions who understand organizational context and are motivated to share insights has proven effective. This model integrates people, process, and product to create a reinforcing cycle of adoption. Ona's strategy involves large-scale presentations to generate interest, followed by targeted workshops to foster power users who become key in spreading AI adoption. The platform's features, like projects that simplify setup and automations that save time, help scale adoption efficiently. Security measures ensure safe scaling, essential for industries with stringent regulations. Successful adoption is characterized by exponential user growth and increased engagement, driven by a partnership approach that adapts to evolving needs.