AI
Individual Productivity Does Not Scale to Team Alignment
Coding agents running in parallel on one person’s machine looks like peak productivity. The promise, one developer with a fleet of agents doing the work of an entire team, assumes software is made by one person. It is not.
“It is a team sport” — Maggie Appleton
More individual output does not solve problems that require communication and coordination. Implementation is getting faster and cheaper, which shifts the real bottleneck to agreeing on what to build. That requires the whole team.
How Agentic Speed Destroys Pre-Implementation Checkpoints
The old development process was slow enough that alignment happened naturally. Draft PR comments, Slack threads, and standups gave teams time to discuss details before code was written. Agents collapsed that window. The time between logging an issue and an agent opening a PR is now a couple of minutes.
Most coding agents have a local plan mode that is completely unshared with other people. All alignment weight falls on the pull request, after the implementation, when it is too late. Features no one asked for, feedback that scraps finished work, merge conflicts from agents touching the same files.
ACE: A Shared Workspace for Humans and Coding Agents
ACE (Agent Collaboration Environment) is GitHub Next’s research prototype, entering technical preview with a few thousand users. Each session is a multiplayer chat backed by a micro VM, a sandboxed cloud computer on its own Git branch. Teammates join to see the full prompting history, run terminal commands, and share a live preview. Nobody stashes Git changes or pulls a new branch to review someone’s work.
The ACE multiplayer session demo (08:47) shows Nate prompting the shared agent inside Maggie’s session. The collaborative plan editing demo (12:35) shows teammates editing an agent-written plan with live cursors before any code is written.
Context at Scale and the Case for Craftsmanship
ACE’s dashboard summarizes what teammates shipped, because co-workers are now pushing five features a day instead of half of one. ACE surfaces names and changes: Nate shipped a lobby channel, David fixed access token issues. It also prompts you to resume unfinished work at the start of the week.
“quality becomes the new differentiator.” — Maggie Appleton
Speed without alignment produces more low-quality software. Agents should buy time for user research and architecture decisions, not just faster shipping. That requires fewer things done better, which means strong alignment from the start.
Notable Quotes
It is a team sport Maggie Appleton · ▶ Watch (1:31)
should we build it? Maggie Appleton · ▶ Watch (2:05)
quality becomes the new differentiator. Maggie Appleton · ▶ Watch (16:03)
Key Takeaways
- Single-player coding agents scale individual output but worsen the team coordination problem.
- Agentic speed collapses pre-build alignment; all checkpoints pile onto the PR, which was never designed for that.
- ACE puts teammates and agents in a shared micro VM so planning and building happen together, not sequentially.