惯性聚合 高效追踪和阅读你感兴趣的博客、新闻、科技资讯
阅读原文 在惯性聚合中打开

推荐订阅源

Google DeepMind News
Google DeepMind News
Microsoft Security Blog
Microsoft Security Blog
B
Blog
Martin Fowler
Martin Fowler
WordPress大学
WordPress大学
爱范儿
爱范儿
博客园_首页
博客园 - 聂微东
量子位
V
Visual Studio Blog
aimingoo的专栏
aimingoo的专栏
T
The Blog of Author Tim Ferriss
J
Java Code Geeks
小众软件
小众软件
大猫的无限游戏
大猫的无限游戏
OSCHINA 社区最新新闻
OSCHINA 社区最新新闻
N
Netflix TechBlog - Medium
F
Fortinet All Blogs
The Cloudflare Blog
T
Tailwind CSS Blog
G
Google Developers Blog
奇客Solidot–传递最新科技情报
奇客Solidot–传递最新科技情报
月光博客
月光博客
腾讯CDC

DEV Community

Authentication Security Deep Dive: From Brute Force to Salted Hashing (With Java Examples) Why AI Systems Don’t Fail — They Drift Spilling beans for how i learn for exam😁"Reinforcement Learning Cheat Sheet" I Replaced Chrome with Safari for AI Browser Automation. Here's What Broke (and What Finally Worked) How Python Borrows Other People's Work The $40 Architecture: Processing 1 Billion API Requests with 99.99% Uptime Vibe Coding: A Workflow Guide (From Zero to SaaS) Most webhook security guides protect the wrong side. The scary part is delivery. Headless CMS for TanStack Start: Build a Blog with Cosmic EU Age Verification App "Hacked in 2 Minutes" — What Actually Happened Comfy Cloud’s delete function does not actually remove files Running AI Models on GPU Cloud Servers: A Beginner Guide Event-driven media intelligence with AWS Step Functions and Bedrock I scored 500 AI prompts across 8 quality dimensions — here's what broke How to Call Google Gemini API from Next.js (Free Tier, No Backend Needed) The Portal Protocol: Reclaiming Human Connection in the Age of AI How to Fix Your Team's Scattered Knowledge Problem With a Self-Hosted Forum Intro to tc Cloud Functors: A Graph-First Mental Model for the Modern Cloud Designing Multi-Tenant Backends With Both Ownership and Team Access I Built a Neumorphic CSS Library with 77+ Components — Here's What I Learned PostgreSQL Performance Optimization: Why Connection Pooling Is Critical at Scale Cómo construí un SaaS multi-rubro para gestionar expensas en Argentina con FastAPI + Vue 3 🚀 I Built an Ethical Hacking Scanner Tool – Open Source Project I Replaced /usage and /context in Claude Code With a Single Statusline A Pythonic Way to Handle Emails (IMAP/SMTP) with Auto-Discovery and AI-Ready Design I Collected 8.9 Million Polymarket Price Points — Here's What I Found About How Markets Really Move EcoTrack AI — Carbon Footprint Tracker & Dashboard Everyone's Using AI. No One Agrees How. 5 self-hosted ebook managers worth trying in 2026 Building Your First AI Agent with LangChain: From Chatbot to Autonomous Assistant
OpenAI Codex Sites: the launch layer arrives — inside you...
LI QINYU · 2026-06-17 · via DEV Community

LI QINYU

When the company that ships the best coding agents also ships "prompt → hosted web app," that is not a competitor showing up. That is the strongest validation yet that agents need a launch layer.

OpenAI's Codex Sites lets an agent turn a prompt into a running, hosted web app — login-gated, with a real database underneath — without you touching a deploy pipeline. That is the "deploy" half of launching an agent-built app, and OpenAI solved it cleanly, inside ChatGPT.

But launching a product is two halves. Deploy is one. Public and paid is the other — a public URL anyone can open, public signup, and a way for the people who use your app to pay for it. Codex Sites stops at the first half. It has no public links, no custom domains, and no payments at all. That second half is exactly what SettleMesh does.

What Codex Sites actually is

Codex Sites went into preview on 2026-06-02. You describe a site or app, the agent builds it, and OpenAI hosts it. Under the hood it runs on Cloudflare Workers, with D1 for the database and R2 for storage. For internal tools — a dashboard for your team, an ops console, a quick data viewer your coworkers need to log into — this is genuinely good. The friction from "agent wrote it" to "it's live and my team can use it" basically disappears. The constraint is who can reach it, and what they can do once they're there.

Availability. Codex Sites is available in ChatGPT Business and Enterprise workspaces (Business on by default; Enterprise via admin RBAC). Plus, Pro, Free, and API tiers don't have it. It's locked inside your company workspace: to use Codex Sites, and to reach what it builds, you generally have to be inside that workspace.

Sharing. A Codex Site can be shared three ways — with admins, with everyone in the workspace, or with named members. All three are scoped to your workspace. There is no "anyone with the link" public mode. That's the line between an internal tool and a product. A product is something a stranger can find, open, and use.

End-user login. Because access is workspace-scoped, the person logging in has to be a workspace member or come through your pre-configured IdP. There's no open public signup the way a normal product has.

Payments. None. Codex Sites has no payment, billing, or monetization capability of any kind. There is no path from "users use this" to "users pay for this."

The fact table (verified June 2026)

Dimension Codex Sites (preview) SettleMesh
Availability ChatGPT Business + Enterprise only (not Plus / Pro / Free / API) Everyone — install the CLI from npm and go
Sharing Three modes, all scoped to the workspace. No "anyone with the link" public mode Public URL, reachable by anyone
End-user login Must be a workspace member or provisioned via your IdP Public signup and login, out of the box
Custom domains None (OpenAI-owned addresses) Platform subdomain out of the box (custom domains on the roadmap)
Payments / billing None Usage billing + user wallet + end-user-pays + hosted checkout
Stack Cloudflare-Worker-compatible ES modules; D1 + R2 Static / SPA / container (any language) + managed DB and storage

Read the table as a story, not a scorecard. OpenAI built a fast, clean path to a hosted, login-gated app for your coworkers. SettleMesh builds the path to a public, paid product for your users. Both are launch layers. They aim at different halves of "launch."

The half that's still wide open

After Codex Sites, an agent can produce a hosted app your workspace can log into. What it still can't do, by design:

  • Give that app a public URL a stranger can open.
  • Let that stranger sign up and log in without being added to your org.
  • Charge that stranger anything — there is no payments layer to charge with.

That last one is where most agent-built apps quietly die. Your AI app burns money on every API call — charge your users for it, per use, without building billing. That's metered billing plus end-user-pays: the end user who runs the expensive call is the one who pays for it.

Here's the four-line version of the other half:

npm i -g settlemesh
settlemesh login
settlemesh deploy
# live: public app with login, a managed database, and usage billing

One command takes an agent-written app — built with any agent, Codex included — and turns it into a public product. Public URL. Public signup and login. Usage-based billing. End-user payments through hosted checkout. Aev is SettleMesh's prepaid credit unit (1 USD = 100 Aev, funded via Stripe): users top up a wallet, and every metered call quotes its cost before it runs and settles against the wallet after.

Who actually lets your AI app charge users

Codex Sites is the newest entrant, but it's not the only tool that gets you most of the way and stops short of payments. The question that matters — can the people who use your app pay you?

Product Publish End-user login Payments Key limitation
Codex Sites Inside workspace Workspace / IdP None Business/Enterprise preview, no public links
Lovable Free to publish Built in (Supabase) Subscriptions + one-time, gated to Pro+; no usage billing Custom domains on Pro+
Bolt.new Free (with badge) Via Supabase Stripe integration; no usage billing Billing logic lives in generated code
Vercel v0 Free to publish No native auth Stripe via generated code Two layers of billing (v0 + Vercel)
Replit Free, 1 app, sleeps after 30 days Yes Integrated Realistically needs Core at ~$20-25/mo
SettleMesh Free tier is enough to ship Built-in public signup Hosted usage billing + wallet + end-user-pays The money path is managed, not generated code

Where the builders offer payments at all, they hand you a Stripe integration or generate billing code, and they mean subscriptions or one-time charges. None give you usage billing — charge per call, per video, per request — as a managed layer. That's the model an AI app actually needs, because the cost an AI app incurs is per use, not per month.

This isn't a roadmap promise

The SettleMesh money paths are already running in production: discrete cost-plus markup (m between 1.0 and 1.5), the unified per-user Aev wallet, end-user-pays via the X-Settle-Payer header, nested billing, and the owner revenue split — with the end-user-pays charge flow verified on a live app. The part that's genuinely a moving target is Codex Sites itself — it's in preview.

The bigger picture

OpenAI shipping Codex Sites tells you the industry now agrees on the shape of the problem: agents build apps fast, and the bottleneck moved downstream to launching them. They solved the inside-the-workspace half. The public-and-paid half is open, and it's the half that turns an agent's output into a product with users and revenue. And it composes: today's deployed app is tomorrow's composable API — every paid app on the mesh can become a capability the next agent calls and pays for.

Try it at settlemesh.io — build with any agent (Claude Code, Codex, Hermes, OpenClaw, Cursor), then settlemesh deploy for the public, paid half.

Build with any agent. Launch with SettleMesh.


SettleMesh is a product of StructureIntelligence Inc. — settlemesh.io · github.com/StructureIntelligence. Aev is SettleMesh's prepaid credit unit: 1 USD = 100 Aev, funded via Stripe.