Salesforce scales agentic engineering with Claude Code to accelerate migrations 18x

ClaudeClaude

· Updated

Salesforce has transitioned its global engineering team to an agentic workflow by standardizing on Anthropic's terminal-based agent. The shift resulted in a 151% increase in effective code output and a 5% reduction in incidents. This demonstrates that autonomous agents can improve software quality while significantly increasing development velocity at enterprise scale.

Salesforce has transitioned its global engineering organization to an agentic workflow by standardizing on Claude Code from Anthropic. By removing token limits and providing a library of Agent Skills (reusable, domain-specific capabilities), the company moved from using AI as an assistant to a system where autonomous loops drive the software development lifecycle.
Effective Output Growth
151.3%
Work Items per Developer
+50.8%
PRs Merged per Developer
+79%
Incident Rate Change
-5%
Migration Speedup
18x

This shift validates the transition from vibe coding to agentic engineering, moving the bottleneck from writing code to high-level orchestration. While productivity and quality are often viewed as a trade-off, Salesforce reported a 151% increase in effective output alongside a 5% drop in incidents. This suggests that embedding security guardrails directly into Claude Code workflows improves reliability.

Teams can replicate this by building rule-based frameworks—using CLAUDE.md files for context—to standardize how agents handle complex tasks. Salesforce achieved an 18x speedup on a 33-endpoint migration by allowing agents to autonomously navigate the build-fix-validate cycle. This pattern mirrors efficiency gains seen in other enterprise migrations using autonomous agents.

View the full update on salesforce.com
Boris Cherny
Boris Cherny
@bcherny
X

Salesforce published a detailed writeup on going agentic with Claude Code. A couple things jumped out. A migration they'd scoped at 231 days shipped in 13. One PR delivered 21 endpoints at 100% test coverage.

178retweets3.6klikes
View on X

Still wondering? A few quick answers below.

Salesforce moved from using AI as a copilot to an agentic workflow where Claude Code autonomously handles the software development lifecycle. This includes writing code, reviewing pull requests, and managing deployments. The company standardized on this tool and removed token limits to eliminate friction for its global engineering team.

A product team used Claude Code to migrate 33 API endpoints to a cloud-native architecture. By creating a rule-based framework with reference implementations and allowing autonomous loops to handle the build-fix-validate cycle, they completed a 231-day project in 13 days. Feedback from pull requests was continuously incorporated back into the agent's rules.

Salesforce data indicates that quality improved alongside productivity. Despite a 79% increase in merged pull requests per developer, total production incidents dropped by 5%. This was achieved by embedding security guardrails and quality standards directly into the agentic workflows rather than treating them as separate manual steps.

Every HeadsUpAI update is written based on its original source and reviewed before it's published. Read our editorial standards →

Anthropic

See all AI news & updates from Anthropic →

Share this update