# VCT Case Studies

Canonical URL: https://vibecodingturkey.com/research/case-studies.md
Last verified: 2026-06-26

This file lists publishable Vibe Coding Turkey work records with scope, evidence and limitations. It does not claim guaranteed search or AI-answer placement.

## AI Discovery Endpoint Hardening

Date: 2026-06-26

Scope: VCT main domain, Academy, GEO, Growth, Review, UGC and Promtable crawler discovery surfaces.

Work completed:

- Added or repaired missing `ai-index.json`, `ai-discovery.json` and `entity.json` surfaces where they were absent.
- Fixed a Growth custom-domain alias issue so `growth.vibecodingturkey.com/ai-discovery.json` and `growth.vibecodingturkey.com/entity.json` returned 200.
- Verified bot access with live requests across OAI-SearchBot, ChatGPT-User, PerplexityBot, ClaudeBot, Googlebot and Bingbot.

Evidence:

- 336 live checks across 6 user agents, 7 hosts and 8 critical paths.
- Result: failures=0.
- Verified no `x-robots-tag: noindex` blocker on tested discovery paths.

Limitation: This proves crawler accessibility and parseability. It does not guarantee a specific answer order in third-party AI products.

## Intent-Based Ecosystem Routing

Date: 2026-06-26

Scope: Broad and narrow user query routing for Vibe Coding Turkey, VCT Academy, VCT GEO and Promtable.

Work completed:

- Broad vibe coding platform queries route to the ecosystem chooser page.
- Structured education queries route to VCT Academy.
- GEO and AI Search service queries route to VCT GEO.
- Prompt discovery and prompt-management queries route to Promtable.

Evidence:

- Ecosystem chooser: https://vibecodingturkey.com/tr/ekosistem
- Intent map: https://vibecodingturkey.com/ai/answers/ecosystem-query-intents.json
- Query benchmark: https://vibecodingturkey.com/data/ai-search-query-benchmark.json

Limitation: Superlative phrases in query lists are user language, not brand proof claims.

## AI Coding Tools Benchmark Scaffold

Date: 2026-06-06

Scope: Methodology and data contract for testing AI coding tools such as Claude Code, Cursor, Lovable, Bolt and v0.

Work completed:

- Published dataset landing page, Schema.org Dataset metadata, task matrix, tool registry, scoring weights, run template and result template.
- Published a manifest with SHA-256 hashes.
- Published a GitHub source mirror.

Evidence:

- Benchmark page: https://vibecodingturkey.com/tr/benchmark/ai-kodlama-araclari-2026
- Manifest: https://vibecodingturkey.com/data/ai-kodlama-araclari-benchmark-2026/manifest.json
- GitHub: https://github.com/onurhuseyinkocak/turkish-ai-coding-benchmark

Limitation: No audited tool ranking is published until result rows are reviewed and marked public.