Our principles.

A product gets shipped only if we expect to be running it in ten years. The first question on any new build is "are we willing to maintain this if it never grows past a small customer base?" If the answer is no, we do not ship it. If the answer is yes, we accept that growth might be slow and that is fine.

We started building Zawjni in 2017 and ran the early version with hundreds of beta members for years while we figured out the matchmaking logic, the privacy controls, and the anti-spam guardrails. The public launch came in 2024 — and the platform now serves 33,000+ members and is still actively maintained. A growth-led playbook would have either killed the early version when growth was flat or scaled the public version aggressively after launch. We did neither, because the people who joined in beta-year two need the platform to still be there in year ten. That gradient — patience before launch, patience after — is the entire point.

The opposite of this rule is the "ship to test, kill if it does not hit hockey-stick" model that VC-funded SaaS uses. It is rational for that capital structure. It is wrong for ours.

Jorbox has never raised venture capital. Has never entertained an acquisition offer. Has never granted equity to outside parties. The company is funded entirely by paying customers — every brand pays its own bills, and the company runs cash-flow-positive month over month.

In 2024 a buyer approached about QRLynx. The offer was meaningful (an exit at that point would have been life-changing for the founder). We declined. The reason was the rule: an acquisition would force a growth curve someone else owns, and the entire business is designed not to have one. Eight months later QRLynx had quietly grown anyway, on its own terms.

The opposite is the dependency-on-future-funding model where every quarter has to outperform to justify the next round. It works for some categories. The trade-off is that the company answers to the next investor, not to the current customer. We picked customer.

Every layer of the product is built or operated by the same operating team. No outsourced hosting provider whose decisions we cannot reverse. No content agency writing about products they do not use. No support call center reading from a script someone else wrote. The cost is that capacity is finite. The benefit is that nothing is anyone else's problem.

In 2015, three years in, we built our own hosting platform instead of moving to a managed provider. The build took six months and was probably twice as expensive as buying. Eleven years later, every brand we operate runs on that infrastructure. We have never had to migrate, never been beholden to someone else's pricing changes, never lost a feature because a vendor sunset it. The compounding return on owning the stack runs for the life of the company.

The opposite is the modern composable-stack model — Vercel for hosting, Auth0 for auth, Algolia for search, Stripe for billing, Datadog for monitoring. It ships faster initially. The compounding cost is dependence on five companies whose pricing, terms, and existence you do not control. Every one of them could change their model and force a migration. We absorbed the upfront cost to avoid that.

Marketing copy is the same prose we would use in an internal Slack message. Support replies are the same prose. Docs are the same prose. The rule is that any sentence containing the words "leverage," "ecosystem," "robust," "delve into," or "in today's fast-paced" gets rewritten. We do not use marketing-AI to generate copy. We do not enforce "brand voice" in the corporate sense. Plain English carries further.

When we shipped QRLynx in 2024, the original product page had been through three rewrites by a freelance copywriter we hired. The third version was technically correct and read like every other QR generator site. We threw it out and wrote 1,500 words ourselves in one sitting. That version of the page is still up. It outranks the copywriter version in every category we measure.

The opposite is the "professional marketing voice" — passive, hedged, conversion-optimized. It does not lie. It also does not say anything. Plain language is harder because it forces specificity. Most companies pick the easier route.