John HiemstraCreativeAI that does real work. You still sign off.
I'm John Hiemstra, a hybrid problem solver: a product engineer and owner building an agentic software development practice to solve business challenges across product, content and system development.
His disciplined approach to brand and UX consistency helped elevate our product offerings at a critical time in our company’s evolution.
How I work
Behind enterprise lines, I've designed product workflows and built shared component libraries used by multiple applications; shipped framework migrations and a platform built for 500,000 concurrent users; led a $40M-a-year legacy-to-cloud move with no customer disruption; and built features working directly with clients, in work used by PwC, CVS, Chevron and the FBI. At a healthcare software company I founded the agentic SDLC practice: five AI-first tools in eight months.
I've built an agentic harness that streams work for rapid iteration on the agile, stream-aligned principles I learned shipping software with teams of talented engineers and product people. I trained my agents in that model: each holds a role in the stream, and every handoff passes a gate. I've always been motivated by delivering products users can trust, because our teams had a process we could trust, and trust to improve when it was found wrong.
Before software, picture framing, woodworking and scenic painting taught me the rule I still work by: measure the job, fit it to the person, stand behind the work. I still work with my hands away from the computer: gardening, woodworking, painting, lighting.
- Start with the job, not the tool. We pick the work that costs you the most time.
- A person approves what goes out. AI drafts; you decide.
- Prove it before you pay for more. Every step is small enough to check.
… appreciate his ability to make the connection between the business and the systems… to understand, design, and implement applications that solve actual business problems.
One I built this year
Our family wellness brand's podcast needed 3–4 hours of work after every episode: show notes, a blog post, social posts, a short clip. Now the transcript goes in and drafts of all of it come out. The host reviews them in 15–20 minutes, for under a dollar of AI per episode.
Nothing publishes until the host approves it.
Code that runs isn't the same as work that's right. I check both.
The posts publish to Tailor, the publishing platform I built this summer to replace WordPress, designed, built, tested and shipped by agents I trained and lead. No plugins to patch, no update cycle to babysit. Pages score 92–100 on Google Lighthouse's mobile test, answer engines can read every page, and it runs a client's site and our family wellness brand's shop, which takes real orders. How Tailor works
able to easily bridge the gap between engineering and product thinking and improved everything he touched
Ways to work together
Three lines of work. Each comes as a project with a defined scope, or as a few days a month on your side.
Start with the job
A project with a defined scope, or a fractional seat on your side.
