How We Use AI | Data Rules & Division of Labor

Practical about the tools. Careful with your data.

“AI-powered” is on every agency's website now. This page is what it means at ours: which parts of the work AI does, which parts a person does, what we never do with your information, and how you can see all of it on your invoice.

The division of labor.

AI does	A person does
Research: competitor sites, search terms, industry sources, summarized fast	Decides which findings matter for your organization
First drafts: pages, posts, emails, scripts, course outlines	Directs the draft, edits it, and takes responsibility for it
Variations: ten headlines, three email versions, a dozen social posts	Chooses, refines, and keeps your voice consistent
Design starting points: layouts, image concepts, video rough cuts	Finalizes design and checks it against your brand
Analysis: traffic, rankings, campaign performance, patterns in data	Interprets the analysis and decides what to do next
Routine operations: scheduling reminders, intake triage, summaries	Configures, monitors, and handles anything the tool isn't sure about

The rule: nothing reaches you without a person with health-system marketing experience having reviewed it. AI makes us faster. It doesn't make us absent.

Faster, cheaper, and better — in that order of proof.

Faster

A ten-page website that once took a hundred agency hours now takes about thirty-five of ours. A quarter of SEO work runs about thirty. You see the hours on the invoice.

Cheaper

Our production rate is $75 an hour because production is faster. When a new tool makes it faster still, your bill goes down — we don't keep the margin.

Better

More variations tested, more research read, more analysis run than any small team could do by hand. The judgment is the same. The raw material is far richer.

Data rules we don't bend.

We do not put patient, donor, or client data into consumer AI tools. Not in a prompt, not in an upload, not “just to summarize.”
Most marketing work needs none of it. Websites, content, campaigns, and analytics can be done well without protected health information, and we scope engagements that way by default.
Where a platform can touch protected information — an intake tool, a patient-communication system, a documentation assistant — we identify that in writing before setup, use only tools whose vendors will sign the appropriate agreements, and configure them to touch the minimum data required.
Your accounts and data stay in your name. We work inside your platforms, not copies of them.
We'll put our practices in writing before you send us a single file. Ask, and we'll send our current AI-use policy.

We replace our own tools every few months. On purpose.

The tools change faster than any agency's process can. So we treat our own team as the test bed: every quarter we evaluate what's shipped, run the promising tools on our own work first, keep what earns its place, and drop what doesn't. Time we spend learning a tool is never billed to a client.

It's the same approach we bring when a practice asks us to evaluate AI tools for its own operations. We've already been through the demos. See Platform Configuration.

Ask us anything about this.

If you have a compliance officer, a skeptical partner, or a question we haven't answered here, bring them to the first call. We'd rather answer it now.

Schedule a consultation