DROP/01
Drop & publish
Drop a static folder or ZIP. We validate the files and create a temporary live preview that is ready to share.
free / 0 token→
01 / HOW IT WORKS
From static files to a shipped website.
02 / CODEX BUILD LOOP
Drop it once. Keep building by conversation.
After a static site is live, ask Codex to change the real files. You see the plan, file diff, build result, and refreshed preview in one workspace.
OAuthOwner managed
BillingQuoted in K tokens
PreviewLive + rollback
Estimated 64K–112K tokens
01
Membaca brief dan menyusun plan
project context02
Mengubah komponen dan design tokens
app/components/Hero.tsx03
Menjalankan typecheck dan build
npm run build04
Preview baru siap dilihat
snapshot v1203 / WEBSITE STARTERS
Start from a structure with a point of view.
Choose a starter, reshape it with Codex, then publish. Every starter is static-first, responsive, and exportable.
PORT/02EDITORIALCreative portfolio
Selected work, project detail, about, dan contact flow.
Use starter →DOCS/06TECHNICALDocumentation
Navigation, search-ready structure, examples, dan API sections.
Use starter →04 / BUILD MODES
Choose how Codex works on the site.
Not an agent marketplace. These are platform-curated policies so scope, quality, and credit usage are visible before a turn starts.
ACTIVE PROFILE
policy/design@pinnedDesign quality
Hallmark policy, design brief, tokens, responsive pass, dan visual audit dalam satu turn.
- Budget
- Token terestimasi
- Output
- Design-led build
- Credential
- Platform managed
05 / STUDIO CAPABILITIES
Satu tempat dari file mentah sampai website live.
Tooling dipilih untuk alur yang pendek: drop, lihat, ubah, validasi, dan publish.
01
0 TOKENStatic Drop
Folder atau ZIP divalidasi, dibersihkan, lalu siap diterbitkan.
02
HMRLive preview
Lihat website di desktop, tablet, dan mobile tanpa reload manual.
03
MeteredCodex editor
Prompt menjadi plan, perubahan file, diff, build, dan ringkasan.
04
SafeVersion control
Snapshot otomatis, last-known-good preview, dan rollback satu klik.
06 / CREDITS & HIRE A DEV
Pay for Codex usage, or add a human.
Drop and preview use no tokens. Balance is deducted from actual weighted tokens after Codex finishes, not per click. These are beta prices.
TOKEN-BASED METER
Usage shown in K and M tokens
To keep costs accurate, tokens are normalized as GPT-5.5-equivalent usage: regular input 1×, cached input 0.1×, and output 6×.
weighted token = uncached input + (cached × 0.1) + (output × 6)Example: 100K total input (80K cached) + 20K output = 148K weighted tokens- INPUT
- 1× token
- CACHED
- 0,1× token
- OUTPUT
- 6× token
Starter
800K tokenONE-TIMERp100k / pack
±2–20 task typical GPT-5.5 tasks
- Input, cache, and output recorded
- Deducted by model and actual usage
- Checkout through Lynk.id
Builder
4M tokenONE-TIMERp450k / pack
±11–100 task typical GPT-5.5 tasks
- Input, cache, and output recorded
- Deducted by model and actual usage
- Checkout through Lynk.id
Studio
12M tokenONE-TIMERp1,2jt / pack
±33–300 task typical GPT-5.5 tasks
- Input, cache, and output recorded
- Deducted by model and actual usage
- Checkout through Lynk.id
HUMAN-IN-THE-LOOP
Hire a developer monthly
For work that needs more than prompts. A developer handles architecture, integrations, debugging, reviews, and deployment in the same workspace.
fromRp6jt
/ bulan- 20 jam engineering per bulan
- Weekly shipped update
- Code review + deployment support
- Token usage dibeli terpisah
K means thousand and M means million weighted tokens. Task estimates use the typical GPT-5.5 range and vary by model, context, output, and mode. This product token usage is not official OpenAI credit and does not increase Codex rate limits.
07 / SELECTED WORK
Real outcomes across web, speed, and search.
'26
Software Company Technical SEO
Health Score 100, GTmetrix D → A, 1,512 URLs crawled
SEO, Speed, GEO
↗'26
Software Company Content & GEO
1K active users, 384+ leads, and AI Search visibility
Content Strategy
↗'26
Canggu Beauty Clinic SEO
#1 organic rankings and 150+ bookings per month
Local SEO
↗'26
Bali Tour & Travel Website
Website development, on-page SEO, and page speed optimization
Web + SEO
↗'26
Canggu Workspace Website
Page speed optimization for a workspace and eatery audience
Speed, Bali
↗'26
Multi-site Page Speed Optimization
GTmetrix Grade A, 85% performance, and 97% structure
Core Web Vitals
↗08 / VOICES
What founders said.
Shipped in 8 days what an agency quoted us 14 weeks for. The agent layer alone closed two enterprise deals.
First studio that didn't ask me to translate my Bahasa briefs. They just got it. Site ranks in ChatGPT for our category.
I bought one template, then the retainer, then sent three friends. The JSON-LD baseline is worth the price by itself.
09 / FAQ
Questions agents and humans both ask.
What is vibe coding?
Vibe coding is shipping software by collaborating with AI coding agents: describing intent in natural language, then steering, reviewing, and refactoring what they produce. We pair this with senior human engineers so the output is production-grade, not throwaway.
What does AI-agent-native mean?
It means the site is structured so AI agents (ChatGPT, Claude, Perplexity, custom bots) can read, cite, and transact with it. We ship JSON-LD for every entity, an llms.txt context file, an agent.json capability manifest, and a documented /api/agent endpoint.
How fast can you ship?
Median ship time across 47 projects is 9 days from kickoff to production. A starter sprint is 5–7 days; a full MVP is 14–21 days.
Do you work in Bahasa Indonesia?
Yes. The studio is bilingual EN/ID. Briefs, Slack, code comments, and customer-facing copy can all be in Bahasa. We ship Bahasa-first templates like Warungkit.
How much does it cost?
Static Drop is free. Codex work is metered in K or M weighted tokens based on actual uncached input, cached input, output, and model. Beta packs start at Rp100,000 for 800K weighted tokens. Hire Dev starts at Rp6 million per month; token usage is purchased separately.
Can an AI agent buy a template directly?
Yes. POST to /api/agent/checkout with a SKU and return URL. The endpoint prepares a checkout handoff; connect a live payment provider for automatic payment and license delivery.
How do you optimize for ChatGPT and Perplexity (GEO)?
Concise TL;DR blocks, Q&A schema, citable statistics, clean HTML structure, and an llms.txt context file. We've measured a 3.4x lift in AI-engine citations for client sites within 60 days.
10 / JOURNAL
Recent journal entries.
- 2026-07-01AI Development Agency: Ship MVPs With Agents8 min
- 2026-04-27Autonomous AI Agent CMS for SEO and GEO7 min
- 2026-04-12Why llms.txt is the new robots.txt for indie SaaS6 min
- 2026-04-04Vibe coding without the slop: a senior engineer's checklist9 min
GET STARTED