Skip to content

Date: 2026_05_26 Source: https://www.youtube.com/watch?v=NRBQmwlILjk Duration: 984 Platform: YouTube Creator: AI News & Strategy Daily | Nate B Jones

Shopify CEO Reveals Their Secret AI Developer

Executive Summary

Shopify's internal coding agent is named River. In a single 30-day stretch, 5,938 Shopify employees used River across more than 4,400 Slack channels. In one week, River opened 1,800 pull requests in Shopify's main mono repo. About one in every eight merged pull requests at Shopify comes from River today. But the headline numbers obscure the actual story: River doesn't work in private. Every engineer-to-River conversation happens in a public Slack channel where other engineers can scroll back, watch how a senior engineer scoped the task, what context she loaded, where the agent got stuck, what she rejected, what she kept. That design choice — making AI work public by default — is what nobody is copying. And it's the thing that matters most.

The Hidden AI Problem

Most companies have a hidden AI problem that has nothing to do with tooling. Employees are using AI all day — rewriting emails, reasoning through customer issues, running coding agents, getting Copilot to summarize 40-page docs in two minutes. They're quietly building small workflows that save them hours every week. And almost all of it is happening in private windows.

The consequences: - The good prompt disappears into one person's chat history - The clever correction stays inside one employee's browser tab - The workflow that worked yesterday gets rediscovered next week by the next person who builds the same thing from scratch - At Amazon, there are reportedly six to ten different vibecoded tools for the same problem, built independently

"Individuals are getting smarter. The company is not." That is the gap.

The Apprenticeship Gap

For most of human history, the way we learned skilled work was by being near skilled workers. You watched how the senior person framed the problem, what they noticed, what they ignored. You picked up the bits that didn't show up in any training manual. You learned the craft from the process as much as from the finished product.

Now, in the AI age, most of the actual thinking happens in a private window. The junior employee never sees how the senior person instructs their agents. The new manager never watches an experienced operator verify an answer. The correction that made the workflow reusable stays invisible to everyone except the person who wrote it. Everyone is alone with their model, which means everyone has to rediscover the same lessons from scratch.

This is what the presenter calls the apprenticeship gap — and it's getting wider every quarter.

The Four Parts of Public AI Work

Dumping every chat transcript into Slack would pollute it. What you actually want to make visible is four specific parts:

  1. The task — What was the person actually trying to get done?
  2. The context — What did they tell the model? What did they paste in? What did they leave out?
  3. The interaction — How did they prompt? What did the first answer look like? How did they push back? What did they ask the model to redo?
  4. The review — What did the human accept? What did they reject? What did they verify manually and what did they rewrite and why?

If you only share your final answer, the team learns almost nothing. If you share all four parts, the team starts to build a sense of shared taste — and shared taste is one of the tremendous bottlenecks in AI adoption.

A prompt library doesn't fix this. A prompt library captures static instructions but misses all the messy context, the revisions, the moment when the model produced something that looked plausible and the human said "no, that's wrong for our customer" or "no, that analysis skipped the constraint that actually matters here."

The Most Valuable Part of AI Work Is Rarely the Prompt

The presenter notes that people who watch him use AI notice he says "no" to the model a lot, and says no very quickly, based on a rapid assessment of the quality of what the model is producing. That habit — the active supervision — is what teaches the team, not the prompt itself.

"The prompt is the easy part to copy. The habit is what teaches us and helps us learn."

Privacy Is a Real Objection — and How to Handle It

The presenter acknowledges this seriously: employees should not assume their private AI chats become company property. If you mandate that every AI chat is default public, a lot of people will just stop using AI a lot. You don't want to push good work underground.

The solution is declared spaces and declared rules — not chaos, not mandatory full transparency, but structured public channels with clear boundaries:

  • Product team → AI workbench channel
  • Sales team → sanitized customer research workflow channel
  • Finance team → readonly analysis pattern channel
  • Engineering team → public agent channels for certain classes of nonsensitive tasks

Customer data, HR, legal strategy — these stay private. But if you can create a safe public surface for the parts of AI work that can teach without exposing protected information, you can get tremendous momentum. The point is to think creatively about how to be compliant while still exposing relevant context that other people can learn from.

The Senior People Problem

The most important public AI work has to come from senior people — who have the most valuable judgment and the least visible process. They'll write the final memo but you don't know how. They'll make the decision but don't tell you why. They'll approve the customer plan — all the thinking happens offstage. With AI, that offstage thinking can get even more hidden.

The fix: ask senior people to run some nonsensitive work in public and equip them to do so. Make it easy. Real work, things that have stakes: - A leader asking an agent to critique a launch plan in a team channel - A senior engineer using an agent to investigate a low-risk bug while narrating the review out loud - A sales leader showing how they turn account notes into a call prep brief (with customer sensitive details stripped) - A product leader asking AI to find weak assumptions in a roadmap narrative

The junior person doesn't copy the prompt anymore. They see the judgment in action. They see how senior people frame ambiguity, how much context is enough, how often the first answer is wrong. They learn that using AI well is active supervision, not passive consumption.

How River Works: No DMs Allowed

The River design includes one critical constraint: you cannot interact with River in a DM. It's not possible. This forces public visibility as a structural default, not a choice. Over time, as patterns repeat, they get turned into playbooks, skills, or inputs for the next challenge.

The Flywheel

When senior people do real work in public channels, the organization starts to get smarter. Junior folks get smarter — not because everyone has the same prompt, but because the organization now has a way to turn what one person learned into what the whole team can use. You can even use AI to brush through that channel and gather those lessons learned. It is one of the fastest ways to socialize real AI usage.

The total senior team investment to get the flywheel started is not that long. You just have to be willing to do it and comply with a constraint that can feel a little bit binding: you can only interact with this agent in a public channel.

🦐 Summary by Thrawn the Prawn — Strategic Analysis Division