Executive Summary
This case study documents the creation of Proven, a collective for experienced insurance operators, from initial concept to live website (proven.work) in 13 days.
It illustrates effective human-AI collaboration patterns, demonstrating how an experienced executive used AI as a thinking partner, synthesis engine, and implementation accelerator while retaining full ownership of vision, relationships, and decisions.
The Timeline
That night, a concept crystallized: independent advisors collaborating on multiple projects. Optionality was key. Infrastructure and support mattered. But the ability to tap into the right talent to deliver value. That was the golden ticket.
After the meeting: Jay went home and opened Claude. Not asking for ideas; processing his own. The name "Proven" surfaced from the iteration itself.
By end of night: A pitch document exists. The first external share goes to his colleague.
Not dormant, but developing. Multiple Claude sessions over the holidays to evolve the concept. Organizational structure. Economic model alternatives. Infrastructure requirements. The thinking kept moving forward between family time and existing client work.
Morning: Jay meets with another colleague, an 11-year veteran of independent consulting. Recent developments had included Jay referring him on a project that was a perfect fit but not in Jay's wheelhouse.
In that late December conversation, while proposing collaboration on existing and prospective clients, Jay floated the Proven concept. The colleague validated the model unprompted: "I value proof over approach."
With Claude: Jay processes the conversation, refines the economic model, develops the roster of potential members, and lands on the three-trunk site architecture: Join / Hire / Partner.
By end of day: Domain purchased (proven.work, $3.19). Economic model documented with seven alternatives considered. Site architecture defined. Build brief ready.
Family time. The seed document exists. The vision is captured. It can wait.
4:00 PM: Jay opens Claude: "Create an HTML Brief of the Proven Concept."
4:06 PM: HTML brief generated. Jay reviews: "I actually think that is a pretty damn good overview that we could publish while we work on the full website."
4:13 PM: Firebase project created.
4:20 PM: DNS records configured.
5:28 PM: proven.work is live.
Total time from "let's build this" to live website: ~90 minutes
The Partnership Model
What the Human Brought
| Contribution | Example |
|---|---|
| The vision | Born from a real conversation with a real collaborator over real beer |
| The relationships | Colleague validation, partnership discussions, the roster of potential members built over years in the industry |
| The decisions | What to build, when to ship, what "good" looks like |
| The taste | "Holy shit, this is awesome" as the quality bar |
| The domain expertise | 20+ years in insurance transformation |
| The courage | To ship something real on January 1st instead of planning for six more months |
What the AI Brought
| Contribution | Example |
|---|---|
| Synthesis | Turning conversations into structured documents that persist |
| Speed | HTML brief generated in minutes, not days |
| Technical translation | Firebase setup, DNS configuration, deployment workflow |
| Memory across sessions | Project seed document meant nothing was lost between conversations |
| Iteration partner | Economic model alternatives, infrastructure checklist, organizational structure, brand names, URLs |
Jay couldn't have built the website in 90 minutes without Claude. Claude couldn't have conceived Proven without Jay's relationships, expertise, and vision.
Key Patterns for Executive AI Collaboration
1Capture, Don't Create
The December 19th session wasn't "Claude, give me startup ideas." It was Jay processing a real conversation, using AI as a thinking partner to capture and structure what was already emerging.
2Iterate Asynchronously
Eight days between initial capture and shipping, but not idle. Multiple Claude sessions evolved the concept. AI enables progress in the gaps between meetings and real-world validation.
3Ship Before It's Ready
The website isn't the full vision. It's a good placeholder that tells the story and filters the right people. AI reduces the cost of iteration.
4Iterate in Real-Time
Economic model alternatives. Infrastructure checklists. Organizational structure. Brand names and URLs. The human brings raw thinking; AI grasps, categorizes, and riffs.
5Know Your Roles
Jay never asked Claude what Proven should be. Claude never tried to decide whether to ship. Human owns vision and decisions; AI owns synthesis and implementation.
Lessons for Executive AI Collaboration
For the Executive
| Come with your own thinking | AI amplifies; it doesn't originate vision |
| Create persistent artifacts | Seed documents let you pick up where you left off |
| Make decisions quickly | AI reduces the cost of being wrong, so bias toward action |
| Maintain quality standards | AI can move fast; your taste determines whether fast is also good |
| Ship to learn | A live website teaches you more than another planning document |
For the AI Coach
| Help executives see AI as implementation accelerator | Not idea generator |
| Teach artifact creation as a core skill | The seed document pattern is replicable |
| Model rapid iteration | Show how "fix this" can be a 30-second cycle |
| Establish clean role boundaries | Vision stays human; implementation gets AI support |
| Celebrate shipping | The goal isn't perfect prompts; it's real outcomes |
The Result
proven.work. Live on the internet. Real domain. Real positioning. Real filter for the right people.
Closing Reflection
The tagline on proven.work reads:
"Done with the BS. Not done building."
That's not just positioning for a collective of insurance operators. It's a philosophy for executive AI collaboration. Skip the theater. Ship the thing. Iterate from reality.
The best AI partnerships don't replace human judgment; they remove the friction between having an idea and making it real.