Europe 2026
Collaborative AI Engineering: One Dev, Two Dozen Agents, Zero Alignment — Maggie Appleton, GitHub
Overview
The talk argues that the current trend of scaling individual developer productivity with AI agents overlooks the inherently collaborative nature of software development. While AI can rapidly implement code, the critical bottleneck has shifted to aligning on *what* to build. Existing tools are ill-equipped for this new paradigm, leading to wasted effort and misaligned projects. The proposed solution involves creating shared, intelligent environments where teams can collaborate on planning and development alongside AI agents.
Who should watch
- AI engineers and developers exploring agentic workflows.
- Product Managers and designers seeking better team alignment.
- Builders evaluating tools for collaborative AI-assisted development.
- Anyone experiencing issues with misaligned AI-generated code or duplicated efforts.
- Teams struggling with the overhead of coordinating multiple AI agents.
Key takeaways
- The promise of individual developer scaling via AI agents is limited because software development is a team sport requiring collective agreement.
- Implementation is becoming a solved problem, making the decision of *what* to build the new bottleneck, with opportunity cost being the real expense.
- Current collaboration tools like Slack and Jira were not designed for agentic development and struggle to handle the volume and nature of AI-generated outputs.
- Alignment has always been a bottleneck, but AI agents significantly increase the cost of misalignment by accelerating implementation without corresponding alignment.
- The context needed for effective decision-making resides in human knowledge (business, product vision, user research) and is not discoverable by agents alone.
- New tools are needed to facilitate constant, shared alignment *before* and *during* implementation, integrating planning, context gathering, and development.
- ACE (Agent Collaboration Environment) is a prototype aiming to provide a shared, multiplayer workspace with integrated AI agents, micro-VMs for isolated work, and real-time collaboration features.
- The goal is to leverage AI to reclaim time for deeper critical thinking, rigorous planning, and building higher-quality, exceptional software rather than just faster, lower-quality output.
Notable quotes
*Software is not made by one person in a vacuum. It is a team sport.*
*Believing individual productivity leads to great software is nine maybe nine women make a baby in one month logic.*
*The hard question is no longer how to build it, it's should we build it?*
Unofficial community note. Prefer the recording for nuance.