Find the workflow your team still completes or covers by hand and name the handoffs, owners, and failure modes.
I help teams make sense of the work before automating it.
Playing Division III basketball taught me to pay attention to the whole team. Who is doing what? Where is the pressure? What happens next? I bring that habit to building AI systems.
Court vision
One shared picture changes the pace.
- Read the floor
- Map
- Call the play
- Build
- Review the film
- Control
See the workflow, owners, handoffs, and opposition before AI acts.
Build the agreed steps, with clear approvals and instructions.
Review results and learn from failures.
How I came to this work.
Court vision, pressure, practice, and review became a way to build AI systems that operators can understand, trust, and own.
Division III basketball made the shared picture visible.
A good possession depends on spacing, roles, timing, and a shared read. An AI workflow needs the same operating clarity before the pace rises.
I played Division III basketball. It taught me to read the whole floor before forcing the next move. A shared picture gives every person a role, a route, and the context to adjust when the defense changes.
That is the model behind CREATE SOMETHING. The Playbook gives operators and approved AI the same view of the workflow. Offense advances approved work. Defense protects decisions, evidence, and recovery.
I grew up between two reference points: a Porsche 930 Turbo and emergency medicine. One taught me that power needs control. The other taught me that when pressure rises, clear protocols matter more than improvisation.
Later, as an equine veterinary technician, I learned what continuity of care, logging, and operating discipline actually feel like in the real world. That stays with me in systems work now.
Today that shows up as calm, transparent, reliable workflow systems: fast where the rules are clear, visible where judgment is required, and recoverable when something goes wrong.
I am a Senior Systems Architect on the Marketplace Team at Webflow. That work keeps the problems concrete: brittle handoffs, unclear ownership, internal tools, onboarding systems, integrations, and platform infrastructure that have to survive real operators.
- Division III basketball
- Porsche 930 Turbo
- Emergency medicine
- Continuity of care
- Webflow systems work
Map the play before AI runs it.
We agree on the steps, who approves them, and how to check the result before giving AI more to do.
Separate safe work from approval-needed work and actions that should be blocked with an inspectable reason.
Connect the tools, write the instructions, and test the work so the client can keep using and changing it.
Surface only the state that needs attention so the human returns for judgment, not constant monitoring.
Advance approved work. Protect operator authority.
We test one task before expanding. The client keeps the workflow plan, rules, results, and software.
Most teams need one workflow that stops creating manual recovery work before they need a larger AI initiative.
Remove repeated steps, unnecessary work, and gaps between tools before automating.
Clients bring the task and the person responsible for it. I help diagnose the problem, build the software, and leave instructions and test results.
Salesforce, HubSpot, Notion, Slack, Procore, Webflow, MCP, and internal systems are useful only when ownership and recovery stay clear.
Every CREATE SOMETHING property improves the next play.
Philosophy, research, experiments, and agency delivery remain distinct while passing useful evidence between one another.
CREATE SOMETHING operates as a connected system. Each property has a distinct job, and each one sharpens the others:
- .ltd — Philosophy and principles
- .io — Research and validated patterns
- .space — Tools and experiments
- .agency — Calm, transparent workflow systems, reliability controls, and enterprise extension (you are here)
Client work informs the research. Research sharpens the operating model. The operating model raises the bar on client work.
I am also building WORKWAY. When clients need fuller system development and onboarding than .agency is meant to carry, I provide a direct referral path to trusted partners, including Half Dozen.
Tell us which task needs too much of your time.
We’ll look at how the work happens today, where AI could help, and which decisions need a person.
- Owner
- Workflow owner
- Authority
- Mapped control states
- Proof
- Safer first path
- State
- review
- 01 / Bring One workflow
The manual operating path that should not need constant rescue.
- 02 / Map Control states
What can run, what needs approval, and what should stop.
- 03 / Leave A safer first path
The operating boundary before a production build starts.