How Salesforce Scaled Engineering AI Adoption Without Stifling Innovation
Paul Constantinides, Anubhav Dhoot and Jackie Chambers de Freitas · 10 November 2026
Read the originalSalesforce's Hyperforce Runtime Platform group spread AI use to about 500 engineers through one shared plugin, with each new skill owned by the team that built it.
About 500 engineers in Salesforce's Hyperforce Runtime Platform organisation, a platform group at the customer relationship management software company, were already trying AI tools on their own. Useful work stayed inside single teams, efforts overlapped, and tight delivery dates left little room for a new programme. An AI steering committee started one shared Claude Code plugin and invited scrum teams to build, deliver and own new skills for their own problems.
- Five contributors built HRPDX, a shared Claude Code plugin that engineers install once and that updates itself with new instructions and skills; 97% of the organisation installed it.
- Scrum teams volunteered for about 30 AI workstreams in one release, the team that built a skill had to deliver and support it, and architects set its quality gates.
- AI Champions carried guidance into scrum teams, brought feedback back to the steering committee and met weekly to show new capabilities.
- The authors report that cycle time fell from 22 days to 11 and sprint delivery rose from 66% to 90% in three months, with the code maintainability score unchanged.
A platform or engineering-enablement team could ship its AI instructions and skills as one self-updating package and make each contributing team own what it adds. This is an inference and is not the authors' claim, and the figures are the company's own account of its programme.
Distribute shared AI agent instructions and skills to every engineer through one plugin that updates itself, instead of asking each engineer to install each skill.
Make the team that adds a skill to the engineers' shared AI coding plugin deliver and support that skill, with architects setting the quality checks it must pass.
Derived by Working Surface from the article; more in the Playbook. Source line: The principle stays consistent: standardize distribution, preserve team choice, and make ownership explicit.
11 November 2026: as AI agents write most of the code, teams are writing their engineers' judgement into shared skills, review rules and testing policies that the agents read every time.