← All work
Technical Build · Proof of Concept

We built an autonomous AI content pipeline — then chose not to run it, and learned more from that call.

End-to-endPipeline built
Claude API (Anthropic)Sanity CMSNext.jsVercelTypeScript
The challenge

Everyone in the AI content space was talking about autonomous content generation. We wanted to know what it actually took to build the real thing — not theoretically, but with working code, real costs, and honest tradeoffs.

What we built

We built ARIA — an end-to-end autonomous blog content pipeline on the Claude API and Sanity CMS. The system generates long-form blog posts, structures them as Sanity documents, and publishes them to the site automatically. No human in the loop.

It works. The architecture is solid, the output is coherent, and the pipeline runs end-to-end without intervention.

Then we looked at the real cost of running it at production volume — and made a deliberate choice not to run it. Human editorial judgment, used strategically, produces better content per dollar than fully autonomous generation at our current volume.

Outcomes
100%
Functional autonomous pipeline
Real
Cost and quality data from production
Clear
Framework for when it makes ROI sense
What this means for you

We can build this for you. And because we've been inside the engine, we can tell you exactly what it will cost to run, where it breaks down at scale, when autonomous generation makes ROI sense, and when it doesn't. That judgment is what you're hiring for — not just the build.

FAQ

Frequently asked questions

What is an autonomous content pipeline?

A system that generates, structures, and publishes content without human intervention — using an LLM (in this case Claude) to write, and a CMS (Sanity) to store and serve the output.

Why did you stop running it?

The token cost of running Claude API at production content volume didn't justify the output quality at our current scale. It's a math problem, not a capability problem — the pipeline works.

When does autonomous content generation make sense?

At high volume (100+ pieces/month), for structured or templated content types, or when human writer costs exceed API costs. We'll help you run the numbers for your situation.

Can you build this for my organization?

Yes. The architecture is reusable. We can adapt it for your CMS, your content types, and your volume targets.

Want to know if autonomous content makes sense for your business?

Tell us what you're trying to build — we respond within one business day.

Let's talk through the tradeoffs