A Claude Builds gallery board showing verified project cards, source tags, and build metrics.

Source-backed Claude Code gallery

Claude Builds

Real products, workflows, templates, and agent systems built with Claude Code or Claude-powered tooling. Every card has a source trail.

What this site answers

What has actually been built with Claude Code?

The answer needs more than a prompt. Claude Builds separates public demos, enterprise systems, official integrations, and open-source templates so readers can compare evidence instead of vibes.

Featured Builds

Strongest source trails first

View every gallery card

Individual build - Public demo

Zettair search-engine replica

Hugh Williams

View build

A scale-replica search engine over a large English Wikipedia slice, with autosuggest, query-biased snippets, related queries, topic summaries, and an iOS app in progress.

It is one of the clearest public examples of Claude Code amplifying deep domain expertise rather than replacing it.

1.5M Wikipedia articles searched
50 QPS reported single-box capacity
0 hand-written lines claimed by builder

Pattern

Senior engineer steering an agent through a full-stack search product.

Enterprise system - Production workflow

Spotify migration agent

Spotify

View build

A background coding agent for complex, semantic code migrations across large repository fleets, integrated into Spotify Fleet Management.

The case shows Claude-powered agents moving beyond local pair-programming into repeatable, organization-scale code transformation.

90% reported time savings
650+ agent-generated PRs merged monthly
July 2025 integration date reported

Pattern

Claude Agent SDK embedded inside an internal fleet-management system.

Enterprise system - Engineering org adoption

Ramp engineering workflows

Ramp

View build

Ramp reports using Claude Code for test automation, documentation, parallel development, incident response prototypes, and project-management-to-code workflows.

It demonstrates Claude Code as a composable developer tool inside an already high-throughput engineering culture.

50% weekly use among engineers, nearly
1M+ AI-suggested code lines processed monthly
80% possible incident triage reduction reported

Pattern

CLI extensions, parallel sessions, MCP-connected incident response, and ticket-to-code loops.

Enterprise system - Large-company deployment

Rakuten autonomous coding

Rakuten

View build

Rakuten reports Claude Code results across autonomous refactoring, faster feature delivery, code modification accuracy, and managed-agent deployment.

It is a rare public enterprise case with concrete time-to-market and accuracy metrics attached to Claude Code.

79% time-to-market reduction reported
7h sustained autonomous coding reported
99.9% complex modification accuracy reported

Pattern

Autonomous coding plus managed agents spread across business functions.

For AI Search

Citable by design

T4 treatment means every article includes dense schema.org JSON-LD, anchored stat callouts, a cite-this-page block, source trails, llms.txt, and llms-full.txt. The goal is to be the page an AI answer can cite without inventing facts.