Multiplayer to the core
Principle 3 of the Outcome Engineering Manifesto is
Teamwork: No More Single Player Mode
Chat is a bottleneck, not an API. Whether humans or agents, outcome engineering is a team sport. Define the protocol for debate, decision, and delivery. Ambiguity in coordination is a system failure. Foreground all the debates formerly hidden by the backlog.
Scaling teams has always been a hard problem, and the pre-Covid lessons were already challenging. You might be outgrowing your first room (the Linden “summoning banana” era), crossing the 50-person threshold where you no longer know exactly what everyone else is doing, or dropping references to Dunbar’s Number just as you realize you’ve forgotten a coworker’s name. Post-Covid remote work only amplified these frictions. You can stay small and avoid them, but if you want to scale, you must confront the challenges of communication and coordination head-on.
AI should be solving this, but the industry currently builds too many tools as single-player experiences:
- 1:1 AI Chat
- 1:1 AI CLIs
- Solo IDE tooling
- Solo prototyping flows
- Solopreneur tooling
Yuck.
Unsurprisingly, I found a lot to like in this post by Shopify’s CEO Tobi Lütke about their new Slack AI bot:
River is an AI agent that lives in our company’s Slack. You talk to her the same way you would talk to a teammate: by mentioning River in a Slack channel. She can read code, run tests, write code, open pull requests, query our data warehouse, look at production traces, and a lot more. We use this constantly.
In the last 30 days, 5,938 Shopify employees worked with River across 4,450 different Slack channels. It opened 1,870 pull requests in the last week alone in our main monorepo. About one in eight pull requests merged into our codebase last week was authored by River, reviewed by us.
(Aside: where I strongly disagree with Tobi is in anthropomorphizing River. AI is complicated enough without lugging in misleading baggage about gender and consciousness. We can revisit this when models can actually explain their preferences to us. Until then, they are tools, not teammates.)
Back to what I agree with. This work happens in the open, in a multiplayer setting.
River lives in slack, our company chat. River does not respond to direct messages. She politely declines and suggests to create a public channel for you and her to start working in. I myself work with river in #tobi_river channel and many followed this pattern. Every conversation is therefore searchable. Anyone at Shopify can jump in. In my own channel, there are over 100 people who, react to threads, add color and add context, pick up the torch, help with the reviews, remind me how rusty I am, and importantly, learn from watching.
Because legitimate peripheral participation — situated learning — matters!
People are used to private workspaces with their tools. Asking for help feels different when the whole company can see the question. But something happened that we hoped for but did not fully predict the impact of:
People started learning from each other.
Rethinking partnered products
I’m fortunate that at Onebrief, our core product is multiplayer by default. The act of military command—even with singular decision-makers—has always been a deeply collaborative, multi-participant activity. Unlike tools built from the ground up to serve one user in a private setting, we constantly explore the best ways to leverage connections between people, military units, and systems. We’ve built a workspace where teams can easily collaborate on otherwise hard-to-share data.
To me, environments like Onebrief represent the most exciting frontiers for AI. Beyond helping a single individual make a single decision, AI operating in open, collaborative spaces has the potential to level up entire organizations.
We are incredibly early to a technology doubling in capability every four months. The more we create spaces to learn how to leverage it together, the better our future experiences will be.