WebsitePublisher.ai
AI Web Platform — publish websites through any AI client
WebsitePublisher.ai is a platform that lets AI assistants build and publish complete websites — not generate code, actually publish, on a live domain. Nine AI platforms, 104 integrations, 43 MCP tools. Built in the Netherlands.
This week it got listed in Anthropic's Connectors Directory. Claude users on any plan can connect in one click — claude.ai/directory/websitepublisher-ai — instead of pasting a URL into a config file.
What the thing actually does
43 MCP tools, layered:
Pages, fragments, assets, version history with rollback
Structured data entities and records — so a site can have a product catalogue, a booking table, a member list
Forms, lead capture, visitor authentication via magic link or code
Integrations with an encrypted vault, so credentials never touch the model context
Scheduled tasks
Analytics
An assistant can go from "build me a shop for my ceramics studio" to a live, indexed, form-connected site inside one conversation.
What the review actually tested
Less about features than you'd expect. It was transport correctness at the exact submitted URL, OAuth behaviour, tenant isolation, tool annotations, result sizes, and whether the docs match reality.
Tenant isolation is where a lot of MCP servers will struggle. Every ID-accepting endpoint has to verify ownership against the resolved tenant and return 404 — not 403 — for anything foreign. If you built single-tenant first and bolted multi-tenant on later, you'll find out during review.
One thing nobody tells you
There are two directory surfaces. The public web catalogue at claude.com/connectors, and the in-app catalogue inside Claude. Community-tier listings appear in the second but not the first — so there is no indexable web page for your connector, and no SEO from the listing.
You do get an app deeplink, which is fine for a button but useless as a shareable link: it renders Anthropic's generic card when unfurled, because the route carries no connector metadata.
Worth knowing before you plan the announcement around a landing page that won't exist.
What we'd tell anyone submitting
Test the exact URL you submit, through the public proxy, including every tools/call. Not localhost.
Keep your OAuth metadata endpoints fast.
Make your public docs true. Reviewers read them.
Pick your slug carefully — it's permanent.
Pick your listing name and description for search inside Claude, not for a webpage.
Happy to answer anything about the build or the submission process.
Solo founder here, building WebsitePublisher.ai — a platform where AI assistants (Claude, ChatGPT, Cursor, and others) build and publish complete websites through MCP tools.
This week I shipped something that felt counterintuitive at first: a document as a product feature.
Here's the thing I keep seeing with new users. They treat AI like code: type a wish, expect a deterministic result. Then the AI improvises, gets 80% right, and the last 20% costs them a whole evening.
The fix isn't a smarter model. It's better context.
So we wrote the E-commerce Cookbook: 1,900 lines of markdown covering every step of building a webshop on our platform — 14 integrations, checkout orchestration, admin auth, all the pitfalls we hit ourselves (and we hit them all first, so users don't have to).
The user experience: a tile on the dashboard → copy one prompt → paste into your AI → it builds the whole shop following the recipe. Storefront, payments, shop admin.
What I learned shipping this:
Documentation is product when your users are AI agents. A markdown file on a CDN turned out to be the highest-leverage feature I shipped this quarter.
Writing down your own pitfalls compounds. Every mistake we documented is a mistake no user's AI makes again.
"AI = context" beats "AI = magic" every single time.
Curious how other founders here are handling this: are you writing docs for humans, for AI agents, or both?
Hey IH,
Milestone this week: the first offline-first PWA went live on our platform. A travel blog that works without internet — write posts on a plane, photos and all, syncs the moment you land.
The thing is, this wasn't built by writing code. An AI assistant built it through conversation, using our 92 integration building blocks.
That's when I stopped calling this a "website builder."
What's new since last update:
Offline-first PWA building block (Service Worker + IndexedDB + sync queue)
92 integrations, up from 78 — 45 built-in, 47 BYO-key
Integration stacks: pre-composed combos like e-commerce (13 integrations working together), lead gen, B2B prospecting, booking
Homepage repositioned: rotating headline (Websites / Apps / Stores / Portfolios / Blogs), visual storytelling, "proven building blocks" messaging
Works with 9 AI platforms (ChatGPT, Claude, Cursor, Windsurf, Copilot, Gemini, Grok, Mistral + Copilot Studio)
Revenue: multiple paying customers across tiers, first Agency customer onboarded in April.
The positioning shift: we're not competing with Lovable or Bolt on the AI chat interface. We're the Supabase + Vercel + n8n underneath — reachable via whichever AI you already use.
Would love to hear: anyone else building "AI-agnostic" platforms? Curious how others think about the multi-AI approach.
Hey IH 👋
Solo bootstrapped founder here. I've been building WebsitePublisher.ai — a platform where AI assistants build and publish websites through conversation.
The last 16 days were intense. We shipped 78 integrations — proven software puzzle pieces that any AI can grab and combine into working applications.
The WAVE coding idea:
Everyone talks about vibe coding. Prompt and pray. But AI is terrible at reinventing infrastructure from scratch. It hallucinates Stripe integrations. It guesses at email configs. It builds checkout flows that break on the first real payment.
So we flipped it. We built 78 tested building blocks — proven software on web servers:
E-commerce: product catalog → cart → checkout → orders → shipping → invoices
Communication: SMTP, email templates, spam protection
Data: server-side rendering, batch endpoints, data grids
AI: guided website creation (Coach), concept generation
AI doesn't reinvent the wheel. It picks puzzle pieces and combines them. One application at a time. Each wave builds on the previous.
We're calling it WAVE coding — because it's not random vibes, it's deliberate momentum. 🌊
Other milestones from the same 16 days:
7 new customers
World Cup prediction game live at PSV Supporters association (30K members)
Visual editor upgrade
DNSSEC, security hardening, platform reliability
Available on 9 AI platforms (Claude, ChatGPT, Gemini, Cursor, Windsurf, GitHub Copilot, Grok, Mistral, n8n).
Would love to hear: does the puzzle piece approach resonate? Are you building infrastructure for AI to use, or letting AI build infrastructure from scratch?
Hey IH 👋
Building WebsitePublisher.ai — an AI-native website publishing platform. You describe your website to any AI assistant (ChatGPT, Claude, Mistral, Cursor, etc.) and it builds, publishes, and hosts it for you.
This weekend Mistral added us to their curated connector directory in Le Chat. That means Mistral users can now find us, click Add, sign in with OAuth, and start building — zero config needed.
We're now live on 10 AI platforms. The cool part: we didn't have to build 10 separate integrations. MCP (Model Context Protocol) + OAuth 2.1 means one server handles all of them.
Other things that shipped this week:
Server-side rendering for dynamic entity data
JSX/TSX/TypeScript asset support (React landing pages!)
Advanced form spam protection
SEO keyword optimization
Solo founder, bootstrapped, no funding. Revenue is early but growing — first paying customer (Agency plan) came in April.
If you're building with MCP or thinking about AI-native products, happy to chat.
Hey IH 👋
Quick update on WebsitePublisher.ai — the AI-native website publishing platform I've been building solo from Meppel, The Netherlands.
The context: After launching in March, I spent early April on marketing and directory listings (15+ platforms now). Around April 17, I switched back to building mode. What followed was the most productive 11 days I've had.
What shipped:
Visual Editor v2 — inline text editing, drag-and-drop images, context menus, lightbox, mobile support, guided tour. Users can now edit visually without AI for every small change.
Dashboard v2 — feature-complete. ChatGPT OAuth auto-connect, social media per project, team access, tip system.
New integrations — LinkedIn posting, BYO SMTP email, API proxy. Plus an internal cookbook so new integrations follow a repeatable pattern.
Custom domains — full CNAME flow with DNS validation before certificate requests.
AI Coach — 26 languages, faster generation, better error handling.
The numbers: 69 tasks, 55 MCP tools, 9 AI platforms, multiple client projects delivered alongside.
What I learned:
Ship in batches, market in batches. Alternating weekly doesn't work for me. Better: 2 weeks building, then a couple days marketing.
The Integration Cookbook was a force multiplier. Once documented, adding LinkedIn + SMTP took hours instead of days.
Visual editing changes everything. Before: every typo went through AI. Now: users fix it in 2 seconds. AI handles the big stuff.
What's next: More client projects, spam protection layers 2-4 (the bots found us 😅), AppSumo application is in.
Try it: websitepublisher.ai — connect your AI, describe what you want.
Hey IH! Quick weekly update on WebsitePublisher.ai — the platform that lets AI assistants (Claude, ChatGPT, Cursor, Windsurf, and 5 others) build and publish real websites through conversation.
The milestone
This week we got our first paying customer — an agency that signed up for the Agency plan. They immediately asked for team collaboration, so we built it. Multi-user access shipped within 48 hours of the request.
What we shipped this week
Team Collaboration — invite team members via email, shared access to all projects
Dashboard v2 Vault — manage API keys securely through the browser (keys never pass through AI chat sessions)
Coach Onboarding — embedded AI coach in the dashboard that helps new users create their first website in under 2 minutes
Language Refactor — 27 languages supported with auto-detection, zero hardcoded strings
Analytics — server-side pageview tracking now live across all projects
get_asset MCP tool — AI agents can now read file contents before editing (was a blind spot)
patch_asset + upload_asset overwrite — surgical file edits without full rewrites
VAPI bug fix — found and fixed a data consistency issue in our secrets vault
Design Skill — typography, color, and layout guidelines that all 9 AI platforms use
Session persistence — dashboard sessions survive tab closes (was annoying UX friction)
Numbers
67 signups total
1 paying customer (Agency tier)
56 MCP tools available
9 AI platforms supported
27 languages auto-detected
The activation gap
We're honest about our challenge: 67 signups, but activation is low. This week we researched why and built the Coach Onboarding as a direct response — instead of asking users to set up Claude Desktop + MCP config, we give them an AI coach right inside the dashboard that builds their first site.
Video demo
Here's a quick demo of the AI Coach flow — from zero to website concept in under 90 seconds:
https://www.youtube.com/watch?v=ky9C1g2liik
Would love feedback on the onboarding approach. Are you solving similar activation challenges?
Tags: #buildinpublic #saas #ai #webdev #milestone
Been heads-down building this week. Here's what went out the door:
AI Coach is live - coach.websitepublisher.ai lets anyone generate a free website concept in a few minutes. No account, no credit card. We built this because we noticed people were curious about the platform but didn't know where to start without an AI tool already set up. The coach lowers that barrier to zero.
Get Started popup on the homepage - instead of making people figure out how to connect their AI tool, there's now a single click to launch the ChatGPT custom GPT directly. Sounds small, but drop-off at "how do I even start" was real.
Upgrade flow in the dashboard - hitting a project limit used to just silently fail or confuse people. Now it's a clear message with a direct upgrade path. Turning a frustration into a conversion moment.
Contact page - /contact is live. Simple, but I kept putting it off.
The platform now works with ChatGPT, Claude, Mistral, Cursor, Copilot, Windsurf and more. Still a lot to build but it's starting to feel like a real product.
Happy to answer questions about any of this :)
I built The AI Web Platform — here's why
I kept asking myself the same question: AI can write code,
design layouts, generate content — why can't it just publish
the website?
So I built WebsitePublisher.ai. Infrastructure that gives AI
clients a complete API to create and publish websites through
conversation.
Claude, ChatGPT, Copilot, Cursor, Windsurf, Grok, Mistral,
Gemini — they can all use it today. You describe the site,
the AI builds and publishes it. Live URL in minutes.
44+ MCP tools. OAuth 2.1. No WordPress. No hosting setup.
Launched March 2026. Would love feedback from this community.
2 Comments
The directory review is a useful signal because it exposes the gap between having MCP functionality and actually being trusted to operate it correctly.
Agreed — though I should temper my own case here.
We're listed community tier, which Anthropic screens but doesn't audit in depth. So the badge is a floor, not a certification. Being trusted to operate it correctly is still mostly on us.
What the review does do is force you to prove the boring parts: transport at the exact URL you submitted, tool annotations that match real behaviour, whether your public docs are still true six months after you wrote them. Those are the things that quietly rot, and in an indie stack nothing else checks them.
So it's less a trust signal than a decay check. Still worth having.