Edit, maintain, optimize
your website with AI.
Describe what you want in plain language. AI agents update content, components, and SEO metadata, and create or modify media assets with GenAI. Next.js and Astro today.
Free to use and self-hostable — no waitlist, no seat count, bring your own LLM keys. You get a preview before anything ships. Early adopters and cofounders welcome. Get in touch.
Capabilities
Built for the modern web.
AI-assisted editing
Agents update real website content: components, metadata, and publishing steps. Not just generated paragraphs.
Live preview
See changes instantly in your own application while agents modify content and layouts in real time.
Works with your stack
Use your existing CMS, DAM, design system, frontend, and AI providers. No proprietary lock-in.
Multi-model
Bring your own models from Anthropic, OpenAI, or Gemini for text, images, and multimodal work. Route different tasks to different models.
Self-hostable
Deploy on your own infrastructure with full control over data, APIs, prompts, and integrations.
Connected workflows
Link editing, approvals, publishing, media, and localization into one workflow you control.
How it works
Describe it. Review it. Ship it.
Nothing reaches production because an agent decided it should. Every change passes through a preview you sign off on.
Describe the change
Say what you want in plain language — no schema spelunking, no ticket for a one-line copy fix.
Agents do the work
Content, components, and SEO metadata are updated in place, and media assets are created or edited with GenAI.
Preview before anything ships
See the result in your own application, with your components, schema, and review rules intact.
Publish on your own terms
Approvals and publishing run through the pipeline you already have, on infrastructure you control.
Architecture
An open layer between AI and your stack.
Avocado Studio sits between AI models and your existing content systems. Instead of replacing your stack, it coordinates work across it.
Next.js and Astro today, with your own components doing the rendering. Extend to the rest of your stack from there.
Websites
Headless CMSs
DAMs
Design systems
Localization tools
Publishing pipelines
- Claude, OpenAI, Gemini
- Avocado Orchestrator
- Next.js, CMS, DAM, Design System
- Preview, Review, Publish
Use cases
What Avocado is built for.
Edit your website with AI
Change website content in plain language while keeping your components, schema, and review rules intact.
Marketing operations
Coordinate publishing workflows across CMS, DAM, localization, and approval systems.
Images and copy together
Generate and edit images with Claude, OpenAI, or Gemini alongside your structured content.
Your own editorial stack
Build AI editing into your own workflows instead of buying a heavyweight enterprise platform.
Developer experience
Built for developers and platform teams.
- API-first architecture
- Git-friendly workflows
- Self-hostable deployment
- Extensible adapters
- Streaming workflows
- Bring your own AI providers
- Works alongside existing CMS platforms
- Open source core
Not another CMS.
Not another AI writing tool.
Avocado Studio is an open AI layer for editing and publishing across the tools you already use.
It helps teams edit websites and publish content with AI agents while keeping their existing stack, infrastructure, and controls intact.
FAQs
Questions worth answering up front.
Is it really open source?
Yes. The core is free and open source, and the source lives on GitHub. There is no seat count and no six-figure platform contract.
Do I have to send my content to someone else’s cloud?
No. Avocado Studio is self-hostable: deploy it on your own infrastructure with full control over data, APIs, prompts, and integrations.
Which AI models does it use?
Whichever you bring. Use your own keys for Anthropic, OpenAI, or Gemini across text, images, and multimodal work, and route different tasks to different models.
Will it replace my CMS?
No — it coordinates work across the stack you already have. Your CMS, DAM, design system, frontend, and AI providers stay where they are.
Can an agent publish something without me seeing it?
No. You get a preview before anything ships, and publishing runs through your own approval and release pipeline.
Which frameworks does it support?
Next.js and Astro today. The Astro integration renders your own .astro components — no React islands, nothing ported. The architecture is an open layer between AI and your stack, so it is not tied to one framework, and more will follow.
AI editing you actually own.
Not rented from a platform, not locked behind a seat count. Run it on your own infrastructure with your own model keys, free. Nothing here is a waiting list — leave an address if you would like a hand onboarding your site, or just start with the docs.