Avocado Studio
AI website ops for Next.js + Astro, in your own components
92 followers
AI website ops for Next.js + Astro, in your own components
92 followers
Hand your website's weekly work to an agent: copy, images, new pages, SEO metadata, translations, site health checks. It works in your own Next.js components, not your CMS database or your code, shows every change on the real page, then publishes a field-level diff to your CMS. Runs on your own Anthropic, OpenAI or Gemini keys. Astro too.
Try the live demo, or run npm create avocado-site@latest. Apache-2.0 on npm. A public research preview: we're looking for early adopters with real sites.
Hi Product Hunt 👋
Avocado is an AI content operations platform and framework for Next.js websites.
🚀 Live Demo: https://avocado-editor.vercel.app/
🧩 Component catalogue used on the demo site: https://avocado-site.vercel.app/catalogue
Most AI-native content tools are bundled inside expensive DXPs, proprietary CMSs, or website builders.
Avocado takes a different approach. Instead of replacing your stack, it adds an AI content layer on top of your existing website.
With Avocado you can:
✅ Edit website content using natural language
✅ Manage pages, metadata, and structured content
✅ Work with real design-system components and content models
✅ Preview AI-generated changes instantly
✅ AI-powered content creation and updates
✅ Connect Contentful, Sanity, Strapi, or custom CMSs
✅ Bring your own OpenAI, Anthropic, or Gemini models
✅ Self-host and keep full control of your data
Think of it as an AI orchestration layer for content operations—not another CMS or website builder. Avocado works with structured content and real website components, enabling AI to understand and modify pages the same way your content and development teams do.
I built Avocado because I believe AI-native content management should be available to every team, not just enterprises with six-figure DXP budgets.
I'd love feedback from:
• Next.js developers
• Content teams
• Agencies
• Headless CMS users
• Composable architecture enthusiasts
What content workflows would you automate with AI?
Thanks for checking out Avocado Studio 🥑 !
Note: The GitHub repository isn't public yet. I'm using this launch to validate the product direction and gather feedback before open-sourcing Avocado Studio under the Apache 2.0 license. If you'd like early access or want to contribute, let me know in the comments.
Congrats! Curious how does Avocado handle content model differences when connecting to multiple CMSs like Contentful and Sanity at the same time?
@crystalmei Thank you ! 🙏 We don't reconcile the CMSs against each other — everything normalizes into one canonical content model (a typed block schema). Each CMS gets a very thin adapter that maps its content types → our blocks on read, and back on publish. So the AI editor only ever touches the canonical shape; adding a CMS is one more adapter into the same model, not an N×N mapping problem. We've got Contentful, Sanity, and Strapi adapters running this way today. The approach is really agnostic of any specific headless CMS, you can even use a plain JSON object store if you wanted to.
content ops for Next.js sites is a sweet spot because the same team is usually shipping marketing pages and product copy from the same repo. curious how Avocado handles the boundary between AI generated drafts and the existing review workflow. does it open a PR like a teammate would, or does it write straight to a CMS layer that the team then approves?
@thenameisarian it doesnt open PRs, it writes into the CMS or any other storage via adapter layer (it would never call the CMS directly or change the code)
Adding an AI content layer on top of an existing Next stack feels way more sane than replacing the whole CMS. BYO models is a nice trust point.
Appreciate it. That's the whole thesis - your CMS, design system, and frontend already work, so Avocado adds an AI orchestration layer on top instead of replacing them. BYO models keeps you in control of keys, prompts, and data too.