PIF Innovate Hackathon 2026 · Riyadh
Local supply does not fail on capability.
It fails on availability.
The Virtual Shared Warehouse — local supply, pooled into one reliable source
Saudi Arabia has built the platforms that answer “who could make this?” — supplier directories, certificates, qualification registries. What no platform can answer is the question a procurement officer actually asks before committing PIF-scale spend to a local vendor: “Is the material on a shelf, in the Kingdom, today — and will it still be there when my order ships?” When that answer is uncertain, the search ends, and the spend leaves the country.
The problem, stated honestly
Problem 01 · Invisible
Supply data is company-level and static
Directories and certificates say what a company is. Nothing says what material exists right now — grade, quantity, certificate, city. A buyer can find a steel company; they cannot find 400 tonnes of certified Grade 60 rebar available this week. So they import.
Problem 02 · Too small alone
No single SME can fill a portfolio-scale order
A PIF portfolio company buys at giga-project scale. Each local vendor alone is too small to bid, so each loses alone — and the aggregate demand that could justify local capacity is never assembled. The order goes to one large foreign supplier instead.
Problem 03 · Too risky
One stockout confirms the import bias
Committing to a small local supplier is a career risk for the buyer: if that one vendor misses one delivery, the project slips. Local sourcing loses not because it is worse, but because it is unbacked — there is no second line of supply behind the promise.
TAWAFUR is that sentence made operational: AI-driven discovery of materials and qualified companies (Layers 1–2), aggregate demand met through composed consortia (Layer 3), and the fulfilment backstop that makes committing local spend safe (Layer 4) — under the MUSAHAMA program’s mandate to deepen local content and localise supply chains at Vision 2030 scale.
The insight: pool it.
Hundreds of small local suppliers, pooled, behave like one large supplier that never stocks out. Materials never move to a shared building — every lot stays on its owner’s rack. The shared warehouse is a ledger: one national, material-level, live view of local supply, with an inter-vendor transfer mechanic underneath it that backs every committed order with a second line of supply. TAWAFUR is named for the missing thing itself: availability.
Five layers, one platform
One national pool. Every local material, visible. Every local order, fulfillable.
The federated inventory graph — the “virtual shared warehouse”
Vendors join with zero integration: CSV upload or a mobile barcode app. Every lot keeps a license-plate ID bound to its grade (SASO/ASTM) and certificate, on the vendor’s own rack. Pooling happens in the ledger, never physically — traceability and quality liability survive intact. Stock carries a freshness stamp; only verified stock is drawable.
Buyer-first discovery
A portfolio-company buyer pastes a messy RFQ, Arabic or English. AI parses it to the material taxonomy — including spec-equivalence: “this certified local grade substitutes your imported spec” — and returns live, quantity-level local supply, not a list of company names.
Consortium assembly
When no single vendor is big enough, the engine composes several into one order — combined price, lead time, and local-content percentage — so many small Saudi suppliers present to the buyer as one large one. Aggregated demand across portfolio companies flows back as a forward signal vendors can invest against.
The liquidity layer — the draw (our core mechanic)
When a committed vendor runs short mid-order, TAWAFUR blind-matches a peer holding the same grade and executes an instant, automated transfer — a sale, not a loan (detail below). The buyer’s order never turns red. Discovery becomes discovery you can trust.
The impact ledger
Every transaction computes SAR shifted from imports to local suppliers, the buyer’s projected local-content score uplift, and Mandatory List coverage — the MUSAHAMA numbers, on screen, per decision.
The draw, precisely — a sale, never a loan
Why it holds: the seller already owns the goods; price is fixed at contracting against a published benchmark; the obligation is observable and dated — nothing depends on the buyer’s private downstream sales. The selling vendor is paid a scarcity-priced markup (utilisation curve), so releasing scarce stock to a competitor is compensated near the option value of keeping it.
The utilisation curve
Scarcity prices itself via the utilisation curve; allocation, settlement and integrity checks are algorithmic.
Worked example
Illustrative figuresThe order
A portfolio company needs 4,000 t certified Grade 60 rebar in 6 weeks. No single local vendor holds it.
Composed
TAWAFUR assembles 3 Riyadh-corridor vendors into one consortium covering 100% of the order.
The shortfall
Week 4: one vendor is 400 t short. Historically, this is where local sourcing dies.
The draw
Blind match to a peer’s pooled stock; murabaha sale executes in minutes; certificates transfer on-ledger.
Delivered
Order ships complete and on time. Dashboard: SAR 11M localised, +0.2 pts local-content score.
Riyadh · Jeddah · Jubail
“Competitors share liquidity, not information.”
Vendors see only banded, aggregate availability per grade and region — never each other’s stock, prices, or identities. Matching is blind and algorithmic; small pools are masked so aggregates never de-anonymise a single depositor. Designed from day one to comply with GAC information-exchange rules; formal competition-law review is scoped alongside Sharia review.
The ecosystem, and the gap
The layer the existing ecosystem is missing — built to plug into it, not compete with it
| Platform | What it does | What it cannot do |
|---|---|---|
| MUSAHAMA Platform | Supplier discovery for 150+ portfolio companies, by local-content performance | Company-level. No material-level stock, no fulfilment guarantee, no inter-vendor liquidity |
| Etimad | Unified government tendering, contracts, payments | Government-only, tender-centric; no private portfolio procurement, no inventory |
| LCGPA score | Measures local content of spend already made | Retrospective — cannot help a buyer find local material before spending |
| Made in Saudi | National brand + directory of certified manufacturers | Static marketing label; no quantities, no availability, no procurement link |
| Jadeer (Monsha'at) | Pre-qualifies SMEs to prove capability | Qualification only — silent on what is in stock and deliverable now |
| Metwafer | One-off marketplace for surplus industrial stock | Ownership transfers per listing; no standing pool, no usage-based settlement |
| Logexa | Marketplace for shared warehouse space | Shares space, not materials — no cross-vendor stock ledger or transfer layer |
| TAWAFUR | The missing operational layer: live material-level supply + consortium fulfilment + inter-vendor liquidity — designed to plug into MUSAHAMA, consuming Etimad / LCGPA / Made-in-Saudi identity as its trust rails. Adoption by PIF is the intended outcome. | |
Against the six criteria
SAR measurably shifted to local suppliers; 20–30% of SME safety-stock capital released; every decision scores its own local-content uplift.
Discovery exists. A pooled, murabaha-settled liquidity layer between competing vendors has no equivalent we could find — in the Kingdom or globally.
Software only: a ledger, a matching engine, an LLM parser. No buildings, no capex. Wedge scope is deliberately narrow.
Adding a sector = a taxonomy branch. Adding a buyer = an account. Adding a region = nodes. No rework, no construction.
Scarcity prices itself via the utilisation curve; allocation, settlement and integrity checks are algorithmic. Substituted imports also cut long-haul freight emissions.
AI where it carries weight: bilingual RFQ parsing, spec-equivalence matching, consortium optimisation, shortfall forecasting, anomaly detection.
The hard questions — answered before they are asked
You are paid for it, near the option value of keeping the stock: the markup rises with pool utilisation, transfers are blind, and pooling returns 20–30% of your safety-stock capital. Contribution is gated — takers must also be givers.
Because there is no lending. Every draw is a sale — title transfers at draw, price fixed at contracting (index + disclosed markup), payment deferred to a dated maturity. Murabaha structure; the word “interest” appears nowhere in the system.
Tiered trust: self-reported stock is visible but only verified stock is drawable — escalating from certificate OCR to third-party spot audit to physically graded intake at Phase-3 hubs.
Because joining costs nothing (visibility tier) and the anchor is named: one portfolio company’s buffer stock seeded on consignment, with aggregated portfolio demand published into the pool from day one.
Wedge first, Kingdom after
Phase 1 · Pilot
One material, one corridor
SASO-certified Grade 60 rebar, Riyadh. Objectively gradable, no shelf life, and exactly what giga-projects buy. Anchor depositor + free visibility tier.
Phase 2 · Deepen
Categories + finance
Polymers, fasteners, packaging. Digital warehouse receipts become financing collateral with a bank / Monsha'at partner — a second reason to list.
Phase 3 · Physical
Hubs where data justifies
3PL-operated shared warehouses per industrial corridor (Riyadh · Jeddah · Jubail), placed where node density proves they pay. The vision, not the dependency.
Scope discipline
Built during the hackathon
The federated ledger over a clearly labelled synthetic dataset (≈200 plausible vendors, 20 commodity SKUs); live LLM parsing of a real RFQ; consortium assembly; the full draw flow — blind match, murabaha contract, on-ledger title transfer; the impact dashboard.
Deliberately out of scope
Procurement workflow, tendering, contracts, ERP integrations, credit underwriting, physical warehousing, and formal Sharia-board and competition-law review — all real, none of them what a hackathon should prove. All figures shown are illustrative.