Wagtail 8.0: A CMS with AI agents, not an AI CMS

Wagtail just shipped a 50+ operation agent API with MCP support, and its most important decision isn't technical — it's philosophical. There are no AI buttons in the admin. No pushed auto-generation to the editor. AI is a tool you install, not a function that gets forced on you. In an ecosystem where every CMS added "AI Magic" as a marketing checkbox, Wagtail went the other direction: hand the developer the pieces and let them decide.

If you read our agentic CMS analysis from 2026, you know seven platforms already shipped agentic capabilities. Wagtail goes further in a different direction: it doesn't package AI — it exposes it as a tool.

What changed in Wagtail 8.0

Wagtail 8.0, released September 8, 2026, brings three concrete changes:

1. v3 API with MCP. Built on Django Ninja and Pydantic, it exposes 50+ admin operations — create pages, edit documents, manage images, review revisions — through a bearer-token-authenticated REST API. The OpenAPI schema is auto-generated. Here's the important part: the API is reachable via MCP (Model Context Protocol), meaning any MCP-compatible AI agent can connect and operate on your content.

2. Included CLI. A command-line interface for the same operations. Content scripting, batch migrations, CI/CD automation with real content.

3. Markdown rich text. Wagtail now supports native Markdown for rich content, reducing friction between editors and agents that work primarily in plain text.

The "opt-in" philosophy is the real differentiator

Most CMS platforms that added AI agents in 2026 did it with the usual product logic: if it's a feature, put it in the user interface. Sanity created Content OS. Salesforce acquired Contentful to wire it into Agentforce. Storyblok added AI assistants. The result: the CMS admin now has a "generate with AI" button that nobody asked for and many editors don't know how to use.

Wagtail does the opposite. Their statement is explicit: "A CMS with AI, not an AI CMS." There is zero AI interface visible unless you install wagtail-ai as an additional package. The agent API exists for developers who need automation, not for editors who need a content editor that works.

Why it matters: the real problem with agentic CMS isn't technical — it's adoption. An editor publishing 3 posts a week doesn't need an agent generating drafts for them. A 20-person team across 5 languages does. Wagtail separates these two needs instead of smashing them into a single interface.

For context: we've already covered how AI crawlers index JavaScript content and how it affects your SEO strategy. Wagtail's API solves a complementary problem: it doesn't just serve content for crawlers — it lets AI agents actively manage it.

Comparison: Wagtail vs Contentful vs Sanity for AI agents

The key difference: in Wagtail, the AI layer is replaceable. If Claude stops being the best option for your use case tomorrow, you swap the agent without touching the CMS. In Contentful, you're locked into Salesforce. In Sanity, you can extend, but their native layer remains theirs.

What it looks like in practice

A content team migrating to Wagtail 8.0 can do this:

# Install wagtail-ai (opt-in)
pip install wagtail-ai

# Configure MCP server for external agents
WAGTAIL_MCP_ENABLED = True
WAGTAIL_MCP_AUTH = "bearer-token"

# A Claude or ChatGPT agent can now:
# - List pages pending review
# - Generate drafts in markdown
# - Assign images to documents
# - Run content audits
# - Publish changes with human approval

The flow: the agent connects via MCP, operates on the 50+ available operations, and the human team reviews before publishing. No hidden magic — it's a well-designed API with solid authentication.

The broader context: Contentful, Salesforce, and the future of CMS

Wagtail doesn't operate in a vacuum. On September 1, 2026, Salesforce closed its acquisition of Contentful for $1B–$1.5B. The plan: integrate Contentful as a native content layer for Agentforce — Salesforce's AI agents. This consolidates the "agentic CMS" movement in the direction of large enterprise vendors.

For teams evaluating their content stack in 2026, the question is no longer "should I use a headless CMS?" but "who controls the AI layer?" Wagtail answers: you do. Contentful answers: Salesforce. Sanity answers: we do, but you can extend it.

What this means for your architecture

If you're building with Astro, Next.js, or any modern framework, Wagtail 8.0's API gives you an integration point that didn't exist a month ago:

• For WordPress migrations: Wagtail is Django-based, Python, open source. The migration gap is smaller than going from WordPress to Contentful. If you're evaluating a migration, check our real lessons from migrating WordPress to Astro — the process is more straightforward than you'd expect when you pick the right stack.
• For content automation: The MCP API means your CI/CD pipeline can include AI agents that review, translate, and publish content without touching the admin.
• For composable architectures: Wagtail integrates natively with Astro, Next.js, and any frontend that consumes its REST API. Compare this with our real Astro vs Next.js benchmarks to understand which framework fits your CMS stack better.

At Mintec, we've worked with agentic CMS since Sanity launched Content OS. The difference with Wagtail is that you don't need to buy a product vision — you need a well-built API and a team that knows how to use it. Sometimes that's exactly what a stack needs.

Published by Mintec — web development consulting, composable architecture, and AI-powered content automation.