Saudi Abwab.ai’s $4m seed targets SME lending stack

SME lending in Saudi Arabia still waits weeks for a human to read a file. Abwab.ai sells the layer that is supposed to shrink that wait without putting the loans on its own books. On 3 September 2026, during LEAP in Riyadh, the company disclosed a 4 million dollar seed, about 15 million riyals. Speedinvest led. Middle East Venture Partners joined. PitchBook also lists Stryde71, Wafra II and a prior Google Accelerator slot from December 2025. Two weeks later at Money20/20 Middle East, founder and chief executive Baraa Koshak repeated the raise on stage and said financial institutions had now processed more than 10 billion riyals through the platform. The company’s own site still prints 1 billion riyals plus across 13-plus lenders. Both figures are company claims. Neither came with an audited book or a default vintage. That is the punch. Four million dollars is a small seed next to Tabby’s consumer machine. The interesting number is throughput Abwab does not fund. Banks, NBFIs, fintechs and funds use the stack to originate, underwrite and watch MSME credit. The pitch is API-first, no core replacement, SAMA-ready audit trail, SIMAH and VAT in the decision file. Website metrics claim 70 percent faster decisions, 90 percent lower cost per case and 40 percent fewer defaults. Treat those as marketing until a named lender publishes the same ratios. Names associated with the platform have included Saudi SME Bank, Abdul Latif Jameel Finance, Lendo, Hala Financing, Kafalah and Raqamyah. Presence on a logo wall is not the same as a live mandate. Koshak founded the firm in 2022 in Riyadh. The team story cites people from Hala, Revolut, NVIDIA and local banks. Abwab is explicit that it is not a lender. That matters in a Kingdom where SAMA wants SME credit to rise as a share of bank books and where the financing gap is still talked about in the hundreds of billions of dollars. An infrastructure company can sit under several licences at once. It can also become a single point of model risk if every lender rents the same score. Use of proceeds is straightforward. Grow the product, hire engineers and machine-learning staff in Riyadh, and push further into the GCC. Earlier company notes mentioned Egypt, the UAE and the UK as operating or target maps. A 2025 founder line about processing 100 billion riyals by 2026 was an ambition, not a booked run-rate. The seed will not buy that share of the market. It will buy more integrations. The MENA angle is the mandate, not another BNPL clone. Consumer names such as Tabby and Tamara take the credit risk. Abwab sells decisioning to the institutions that must hit SME quotas without hiring a floor of analysts. That is closer to industrial policy than to a checkout widget. It is also why SAMA alignment and a full audit trail sit in the homepage copy. A model that cannot explain itself in Arabic and English will not survive a supervisory visit. Caveats belong with the 4 million dollars. Processed volume is not originated volume and not Abwab revenue. Jumping from a public 1 billion riyal milestone in 2025 to a 10 billion riyal stage line in 2026 needs a reconciliation the company has not published here. Faster decisions can mean looser decisions. “Fewer defaults” without a cohort and a vintage is a slogan. Seed capital from European and regional VCs does not replace a banking licence if the product ever drifts toward funding the loans itself. If the next twelve months add named production lenders and a default series that holds, Abwab becomes the quiet rail under Saudi SME credit. If the 10 billion riyals stays a keynote number, the seed bought a louder story. The infrastructure test is whether a credit committee will still use the model when the cycle turns.