All Episodes

What It Do: The 90% Rule - Why Finishing Is the Least Fun Part of Building Anything

with Jason Katz · Kindling Solutions

July 6, 202600:33:14Littleton, CO

What It Do: The 90% Rule - Why Finishing Is the Least Fun Part of Building Anything

0:000:00

Show Notes

Jason Katz popped his LCL three weeks ago scrambling out of a jiu-jitsu leg entanglement, and honestly, that injury is the whole episode in miniature. You get excited, you move fast, you stand up into a hold you did not see, and something structural gives.

Funding stage: Bootstrapped. Kindling Solutions is a founder-funded AI systems consultancy with no announced outside capital, growing on client revenue and a soon-to-be 1099 sales team.

In this What It Do check-in, Jason, co-founder of Kindling Solutions and back for another round, and Ryan trade war stories from the two weeks that turned both of their companies from build mode into go mode. Jason drops the concept that should be tattooed on every founder's forearm in 2026: built is not built to scale. Anyone can vibe code something that works in a single session now. He points to reports of vibe coders getting sued after losing company data through open-ended permissions, and he watches CEOs get star-gazed by AI demos, fire people AI cannot actually replace, then quietly rehire them. Kindling's answer is a governance layer, a way to let a client's internal Lovable and Claude Code tinkerers keep building while Kindling governs the builds it does not even touch. Operators first, then engineers. Jason says his team has exited three companies at nine and ten figure outcomes (company-stated, not independently verified), and that operating scar tissue is the product.

Ryan, meanwhile, is in full sales dog mode. He restructured his entire week, calls only on Tuesdays and Thursdays, deep work on Monday, Wednesday, and Friday, and promptly signed two newsletter sponsorship deals in one week: Momentous, the NSF Certified supplement company behind the grass-fed whey and creatine he already takes daily, and Taelor, the AI-plus-human-stylist menswear rental subscription about to give him a wardrobe glow-up, per his daughter's demands. (AI for Founders already featured Taelor's Zoher Karu in The Self-Driving Car of Men's Fashion.)

And the whole thing wraps with the two of them planning a Denver panel with Gabe Anderson and ID345's Danny Newman that may or may not end in a staged brawl. A leopard-print Speedo has been threatened.

Named Frameworks

The 90% Build

Jason's rule for where creative energy actually belongs.

  • Build in mad scientist sprints to 90% completion, where all the creative leverage lives.
  • Accept that the final 10% is administrative standardization, not creation.
  • Hand the polish to people who are wired for it instead of stalling in it.
  • Budget a year to harvest all the derivatives a good system produces.

Built vs Built to Scale

The gap where AI projects quietly turn into lawsuits and layoffs.

  • "It works in one session" and "it works for an organization" are different products.
  • Scale requires information control, data control, permissions, and governance.
  • Map your customer acquisition and fulfillment lifecycle before you automate it.
  • The delta between built and built to scale is where costs explode and wrongful layoffs happen.

The Idea Cooling-Off Period

Ryan's filter for which ideas actually deserve execution.

  • Colorado makes you wait 48 hours to pick up a pistol; ideas deserve the same respect.
  • Capture everything, act on almost nothing immediately.
  • If you still want it after the wait, it earned execution.

The Three-Tier Filtration System

Jason's cascade for surviving inbound chaos.

  • Tier 1: Notion holds all projects, tasks, and deep context.
  • Tier 2: when inbound chaos spikes, distill to a digital notepad.
  • Tier 3: when it gets really crazy, physical paper and a Post-it, valid for two to three days max.
  • Run a hygiene pass after: wipe the paper, wipe the notepad, reorient the root system.
  • Whatever survives to the Post-it is the gold standard, validated by friction.

The Calendar Firewall

Ryan's structural fix for a week eaten by calls.

  • All calls and podcasts stacked on Tuesdays and Thursdays.
  • Monday and Friday for admin deep work, Wednesday for uninterrupted building.
  • Accept the consequence: brutal stacked days in exchange for protected creation days.

Founder Experiment

Install the cooling-off filter. For seven days, no new idea gets executed the day you have it. Capture every idea in one place, Notion, a notes app, whatever you use. Every 48 hours, review the list and promote only the ideas you still feel in your chest to a physical Post-it. At the end of the week, count how many ideas you captured versus how many survived to paper. That ratio is your personal noise-to-signal number, and it will probably humble you.

Glossary

Vibe coding

Using natural language prompts to have AI write software, no traditional engineering required.

Built to scale

Software plus the governance, permissions, security, and process needed for an organization to rely on it.

Governance

The rules controlling who and what, including AI agents, can access data and take actions.

No-gi

Jiu-jitsu training without the traditional uniform, faster and more leg-lock heavy.

LCL

Lateral collateral ligament, the outer knee ligament Jason sprained.

PMF (product-market fit)

The point where the market pulls your product out of you.

ICP

Ideal customer profile, the specific buyer you build and market for.

1099 rep

A contract salesperson paid on performance rather than salaried.

Content derivatives

The many formats, short video, long post, thread, spun from one core idea.

NSF Certified for Sport

Third-party certification that a supplement is tested for banned substances and purity.

Questions This Episode Answers

  • What is the actual difference between "built" and "built to scale," and why does that gap get founders sued or sink their AI projects?
  • My team is vibe coding internal tools. How do I let them keep building without exposing the company to data loss and lawsuits?
  • Should I fire people based on what a single AI demo makes look possible?
  • I have more ideas than time. What system actually filters which ones deserve execution?
  • How do I protect deep work time when sales calls and podcasts keep eating my week?
  • How do podcasts and newsletters actually land sponsors without burning audience trust?
  • What can a 48-hour idea cooling-off period do for a founder who chases everything?

URLs Mentioned in the Episode

Links & Resources