2–3 weeks · fixed fee
Content Engine Blueprint
Every engagement starts here.
Process and workflow mapping
A context and data readiness assessment
Your Voiceprint: tone of voice rules and guidelines ready for AI agents to use
The engine specification and a staged build plan
Tools and tech stack recommendations
8–12 weeks · fixed fee
Content Engine Build
You own everything.
Context architecture
Prompts and skills library
The Voiceprint Loop, wired in and running
Quality gates and the workflow, end to end
Training for your team on how to use the system
Monthly retainer
Maintain and extend
Available after a build. Three-month minimum.
Implementation of new use cases
Monthly quality sampling and engine performance evaluation
Voice-drift checks
Model and tool changes absorbed
Nobody argues with an architect about whether drawings come before construction. Two to three weeks, fixed fee, mostly asynchronous. You leave with the specification and a staged plan whether or not I'm the one who'll build it.
If you already know you want the build, we can run it as one engagement in two phases: the Blueprint first, then the build quoted against the specification within a range I commit to up front.
Engagements are fixed-fee and bounded.
I’ve built content systems in enterprise environments where the interesting tools were banned outright: limited AI model access, no connectors, an approved Microsoft stack and nothing else. That constraint doesn’t make the work impossible. It changes what you build first, and it makes the documentation and the taxonomy matter more.
If your IT function is a committee and your answer to most tools is a no, that is a normal starting condition for many companies, and I’d rather design for it from day one than hand you a plan you can’t get approved.
The method
How I work
1 · Specification
Interviews with the people doing the work, so the process that gets written down is the one that actually runs. A context and data readiness assessment, your Voiceprint extracted, then the engine specification and a staged build plan.
2 · Build
Context architecture, the prompt and skill library, and the end-to-end workflow built on the tech stack we've identified and that you have access to.
3 · Quality gates
A human stays in the loop wherever AI produces something. Where volume makes full review impossible, I implement sampling methods that give you statistical confidence to produce content at scale.
4 · Training
Your team learns to run and extend the engine. If it only works when I’m in the room, it isn’t finished.
5 · Handover
You own it. Every prompt, every workflow, every document.
The problem is that “our voice” usually lives as adjectives. Professional but approachable, authoritative but warm. But adjectives cannot be checked, and an AI model cannot be held to them (and honestly, neither can a human copywriter).
So the first thing I do? Turn your voice into something testable.
Ownership
What you own
Everything I build is yours: the prompts, the workflows, the documentation, the Voiceprint, the evaluation rubrics. You can run it, change it, extend it and take it to someone else without asking me.
I don’t resell software, and I take no commission from any tool I recommend.
Boundaries
What I
do
Content as a service
No article retainers, no volume deals. I build the engine. Your team runs it.
Automated posting
Your brand is too delicate to hand to a scheduler, and I haven’t seen it work.
Work with teams who can’t own it
If nobody internally is going to run this after I leave, it won’t stick.
FAQ