Capstone: draft a cohort-three SBP sandbox application for a hypothetical neobank

Three ways to see it

1. 1

   This capstone is the integration test for the previous seven lessons. Lesson one gave you the charter taxonomy. Lesson two built the remote KYC stack. Lesson three opened the data layer through open-banking APIs. Lesson four made Raast a primitive. Lesson five built credit scoring on alternative data. Lesson six designed the customer-ops AI. Lesson seven taught the sandbox application craft. The job now is to fold all seven into a coherent application for one specific company, called Nukta, that you will draft section by section.

2. 2

   Nukta's positioning, refined through three weeks of customer interviews in Faisalabad and Multan, is this: 'A consent-based mobile bank that lets women running home-based stitching, salon, and tuition businesses see all their money in one place, get paid by Raast under their own name, and qualify for working-capital credit using their JazzCash and utility records.' The single sentence does four things: it names the customer, the unmet job, the rail it sits on, and the new capability that would not exist without the sandbox. If your one-line draft from earlier is missing any of those four, edit it now.

3. 3

   The cohort plan is where most teams over-promise. Nukta's discipline is: maximum 1,500 customers in the test window, geographic limit Faisalabad and Multan only, maximum exposure PKR 50,000 of credit per customer, maximum deposit PKR 200,000 per customer, six-month duration. Onboarding uses the four-stage NADRA Verisys + liveness + match flow from lesson two with a fallback path involving a video-call agent for any borrower whose face match scores below 0.82 or whose CNIC photo is older than seven years. Customer-ops is built on the layered AI from lesson six, with mandatory human review for any credit decision above PKR 25,000 and any dispute marked emotional by the bot.

The why-tree

Try this with Claude

AI-edge prompt: 'I am submitting a regulatory sandbox application to SBP cohort three for Nukta, a Raast-native neobank serving women micro-entrepreneurs in Faisalabad and Multan. Here is my one-page summary [paste], my cohort plan [paste], and my risk register [paste]. Play the SBP committee in panel format with three personas: a sceptical risk officer, a pro-innovation deputy governor, and a consumer protection specialist. Each asks me their hardest question. After my answers, score me out of 30 and tell me whether I would clear the cohort gate.' Iterate until you score above 24.