A Saudi founding team came to us after four years of living with an idea and one failed agency build — deep conviction that property management in the Kingdom was badly underserved, a concept paper that spanned five different businesses, and no product thesis, no engineering team, and no way to tell which part to build first. We built all of it: the strategy and the wedge, the feature set, the brand and design system, the architecture, and the platform underneath.
The result is Shajr — a bilingual (Arabic/English), KSA-hosted portfolio and lease management platform for institutional landlords and property managers, spanning two connected applications: an Operator Console where customers run properties, units, owners, tenants, leases, financials, and service operations, and an Admin Platform where the business itself is run — provisioning, entitlements, billing, support access, and monitoring. Underneath sits a live rent roll that replaces the monthly spreadsheet rebuild, return per building instead of return by account, AI lease abstraction that cites the exact page and clause it read, and a market-intelligence layer built on official Saudi open data.
Shajr is delivered and live. We're now in an active upgrade phase with the founding team — expanding the feature set and onboarding additional pilot customers on top of the platform we shipped.
The founders had genuine market insight but no vehicle for it. Their concept paper described a destination — listings marketplace, property management, facilities management, payments, analytics, government integrations — where each layer is a venture-scale problem on its own, and none of it answered the only question that matters first: who pays for the very first version, and what are they doing without it today? Four years of scope movement and one unsuccessful prototype had produced a vision with no product inside it.
The domain itself was unforgiving. This is not a CRM with Arabic strings bolted on: it is one tightly coupled financial graph — lease to rent schedule to invoice to ledger to owner statement — where a single wrong indexation cap or proration rule silently corrupts every number downstream. It had to be genuinely bilingual, right-to-left and Hijri-native rather than translated after the fact, because the contracts, the statements, and the legal dates are Arabic-first. And the real adoption barrier was never features: every prospect's portfolio lives in spreadsheets today, so onboarding hundreds of leases without corrupting them is the product, not a step before it.
Layered on top was compliance that could not be retrofitted. A multi-tenant platform holding the lease and financial data of competing Saudi property firms had to be built for PDPL, the NCA Essential Cybersecurity Controls, ZATCA invoicing, and in-Kingdom data residency from the first migration — in a system where one row crossing between two tenants is not a bug but an existential event.
In plain terms: we turned an idea that tried to be five companies at once into one focused product a real customer would pay for on day one — then built it so the numbers a landlord sees are always trustworthy, one customer's data can never leak into another's, and every AI suggestion is approved by a person before it touches a lease, an invoice, or a statement.
We started above the code. We repositioned the concept from a comprehensive platform to a defensible wedge — Saudi-compliant portfolio operations for large property managers — pressure-tested the ICP through structured stakeholder interviews, ran the build-versus-buy question honestly, and turned the outcome into a 99-feature PRD with an explicit MVP boundary. Alongside it, we built the brand, the bilingual design system, the information architecture, and the system architecture the platform is still held to today.
Engineering was built guardrails-first. Tenant isolation is enforced twice — in the application and again by PostgreSQL row-level security — and proven by an isolation suite that runs in CI. Every write lands in an append-only, hash-chained audit log, so retroactive edits are detectable rather than merely discouraged. Onboarding runs through a single import engine with a row-level dry run and a reconciliation against the customer's own control totals before anything goes live. Arabic shaping, right-to-left layout, and Hijri dates hold all the way into generated PDFs.
AI was governed before it shipped. A single gateway is the only code allowed to call a model, and each AI feature declares exactly which fields it may send. Lease abstraction, invoice capture, and expiry detection land as suggestions with page and clause citations beside the source document, and a person accepts each one — nothing auto-commits. Weeks of analyst time abstracting a 200-lease portfolio becomes a review measured in hours.
Delivery is AI-orchestrated under senior human judgment. Claude plans and specs, a separate model adversarially reviews every high-risk spec, Codex writes the code, and every change passes three human-led gates — local QA, a CTO-level review, and a second QA pass on staging — before customers see it. More than 200 automated tests back it, and no model's claim that its tests pass is ever accepted without independent re-verification.
In roughly ten weeks from kickoff, a founding team with no technical function went from a concept paper to a running, multi-tenant, bilingual platform — with the positioning, brand, design system, 99-feature product definition, and full architecture built alongside it. The operating core itself went from an empty repository to properties, units, owners, tenants, leases, the rent-schedule engine, financials, reporting, the import engine, AI features, and the market-intelligence layer in under three weeks of build, across 97 reviewed and merged changes.
Because compliance, tenancy, audit, and AI governance were built into the spine rather than added later, the platform did not have to be re-architected to be credible to institutional buyers: the data residency story, the tamper-evident audit trail, and the human-in-the-loop AI posture are properties of the system, not promises about it. Every feature ships inside those guardrails by construction — AI moving fast, senior judgment deciding what's safe to ship.
Shajr is delivered and live today, with the founders holding a defined product, a shipped platform, and a roadmap they can actually sell against. We're now in an active upgrade phase together — onboarding pilot customers and expanding the feature set on top of the platform we shipped.
Want similar results for your business?
In submitting this form, you are agreeing to Fynch’s Privacy Policy.
This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.
Fast. Smart. Always on.
Fynch’s AI powered assistant is designed to help you unlock our full potential
Explore our product audit
Spot product and AI-readiness gaps early with our 80+ parameter audit, powered by McKinsey, Capital One, and RingCentral experience.
