Descriptions:
Nate B Jones analyzes a public post from Shopify CEO Toby Licki revealing detailed usage data for River, Shopify’s internal AI coding agent. The numbers are substantial: over a recent 30-day period, River was used by 5,938 Shopify employees across more than 4,400 Slack channels, opened 1,800 pull requests in a single week, and now accounts for roughly one in eight merged pull requests in the company’s main monorepo.
Beyond the headline metrics, Jones focuses on the design decision that sets Shopify’s approach apart: every engineer’s conversation with River happens in a public Slack channel, making the reasoning process—how senior engineers scope tasks, load context, reject agent output, and iterate—visible to the entire team. He frames the problem most companies face as the ‘apprenticeship gap’: when AI-assisted thinking happens in private browser windows, institutional knowledge about how to work effectively with models never propagates. Junior employees can’t observe expert workflows; effective prompts disappear into individual chat histories; teams rediscover the same lessons quarter after quarter.
The practical framework Jones outlines involves creating declared public Slack channels for non-sensitive AI work—segmented by team function—with explicit boundaries around customer data, HR, and legal. He draws on analogies from manufacturing knowledge transfer and references his own Substack for a deeper breakdown of the four-part interaction pattern (task, context, interaction habit, review standard) that makes AI work learnable and shareable at the organizational level.
📺 Source: AI News & Strategy Daily | Nate B Jones · Published May 26, 2026
🏷️ Format: News Analysis
