Work with me · Option A
AI content systems
Content at scale that doesn’t read like it. I design the whole machine — strategy, tooling, and the team workflow around it — then stay until it runs without me.
Problem
Content is a grind, and scaling it kills quality
You know content works — that’s not the question. The question is how to publish more without every piece becoming a negotiation: briefs that go stale, writers who burn out, review queues that grow faster than the blog.
Bolting AI onto that doesn’t fix it. Raw model output reads like raw model output, and your editors end up rewriting everything anyway. The problem isn’t the tool — it’s that nobody designed the system around it.
What I build
Human-in-the-loop pipelines — the whole machine
Not a prompt library. A production system where AI does the heavy lifting and people do the judging, designed end to end:
-
Strategy
What to publish, for whom, and why — mapped to search demand and your actual expertise, so the pipeline never runs on empty.
-
Tooling
The pipeline itself: models, prompts, data sources and QA gates, wired into the tools your team already uses. [PLACEHOLDER: example stack]
-
Team workflow
Who touches what, when, and what “done” means — editors reviewing at the points where human judgement actually changes the output.
How it works
Four steps, drawn in order
-
Diagnose
We map your current content operation: where quality is made, where time is lost, and what is worth automating at all. [PLACEHOLDER: duration]
-
Design
I draw the system on paper first — pipeline stages, review gates, tone rules — and we agree on the drawing before anything is built.
-
Install
I build the pipeline, wire it into your tools, and train the team on it — on real briefs, not demos.
-
Run
We run it together until the output is consistent, then I hand over the controls. Documentation included; dependency not.
What you get
Deliverables, not decks
- A content strategy mapped to real demand — topics, formats and priorities [PLACEHOLDER: format]
- A working pipeline installed in your tools, with prompts and QA gates documented
- A team workflow: roles, review points, and a definition of done
- An operating manual, so the system survives staff changes
- [PLACEHOLDER: support period] of adjustment support after handover
Fit
Who it’s for — and who it isn’t
For you if
- You have an in-house team (or steady freelancers) who will run the system day to day
- Content is a proven channel, and the bottleneck is volume or consistency
- You want your editors making judgement calls, not typing first drafts
Not for you if
- You want fully automated publishing with no human review
- You need a few one-off articles, not a repeatable operation
- Nobody on your side can own the system after handover
Questions
Asked on most calls
Will the content read like AI wrote it?
Not if the system is doing its job. Human judgement is designed into the pipeline at the points where it changes the output — voice, claims, structure. The model drafts; your people decide.
How long does a build take?
Depends on the size of the operation — typically [PLACEHOLDER: timeline range] from diagnosis to a running pipeline. You’ll have a fixed plan after the Diagnose step.
Do you write the content yourself?
No — your team runs the machine. I design it, install it, and stay through the first production cycles. That’s what makes it stick.
Have a specific problem?
Bring it to a 30-minute call. If I can’t help, I’ll tell you who can.