SaaS Development Company in Toronto

Fintech / Open Banking

A Toronto-based fintech (Series A, $14 million CAD raised) was building an open banking data platform — middleware SaaS that enables Canadian fintechs, lenders, and financial advisors to access bank account data (with consumer consent) from the Big Five banks. Canada's open banking framework was being implemented, and the company needed to be ready. The prototype connected to 2 banks (TD and RBC via screen scraping — a fragile method) and provided basic account and transaction data. Problems: screen scraping instability (bank website changes broke the connection approximately every 2 weeks — causing data outages that destroyed fintech customer trust), limited bank coverage (only 2 of 5 major banks and none of the 30+ smaller banks and credit unions), no consent management (the prototype used customer-provided credentials with no formal consent framework — inadequate for the coming open banking regulations), and poor data quality (screen-scraped transaction data was inconsistent — merchant names, categories, and amounts varied in format across banks, making the data difficult to use for analytics). The company needed 30 fintech clients within 12 months to demonstrate market leadership before open banking regulations solidified.

We rebuilt the platform as a production-grade open banking data aggregator with multi-bank connectivity, standardised data, and consent management. (1) Multi-bank connectivity: moving beyond screen scraping. API connections: where banks offer APIs (TD and RBC have developer programmes), we connected via official API channels — more stable, faster, and compliant. Screen scraping with resilience: for banks without APIs (which is still most Canadian banks), we built resilient screen scraping with: multi-layer parsing (not dependent on specific HTML structure — using semantic understanding that survives website redesigns), automated monitoring (detecting connection breakage within minutes — not days), self-healing (common breakage patterns automatically corrected without engineering intervention), and graceful degradation (when one bank connection fails, the platform continues operating with cached data and clear status communication). Coverage: all 5 Big Five banks + 12 credit unions + 8 smaller banks = 25 financial institutions covering 94% of Canadian consumer accounts. (2) Data standardisation: raw bank data normalised to a consistent format. Transaction enrichment: merchant name cleaning (converting "AMZN MKTP CA" to "Amazon"), category assignment (using ML model trained on 50 million categorised Canadian transactions), and location extraction (identifying the city and province for in-person transactions). Account normalisation: consistent account types (chequing, savings, credit card, investment, mortgage) across all banks — each bank labels these differently. Balance and transaction history: standardised date formats, currency handling (CAD primary, USD where applicable), and pending vs. posted transaction states. (3) Consent management: built for Canada's coming open banking framework. Explicit consent: customers grant consent through a branded consent flow — specifying which accounts, what data, and for how long. Consent dashboard: customers can view, modify, and revoke consent at any time — revocation immediately stops data access. Purpose limitation: each fintech client specifies their data use purpose — the platform enforces purpose limitation, only providing the data elements necessary for the stated purpose. Consent audit trail: every consent grant, modification, and revocation logged — creating the audit trail that open banking regulations will require. (4) Developer experience: making it easy for fintechs to integrate. RESTful API with comprehensive documentation, SDKs for Python, JavaScript, and Ruby, sandbox environment with synthetic data for testing, and webhook notifications for new transactions and balance changes. Onboarding: a fintech can go from signup to first API call in under 2 hours. (5) Security and compliance: SOC 2 Type II from project inception. Data encryption (AES-256 at rest, TLS 1.3 in transit), access controls (API key + OAuth 2.0 authentication), rate limiting (preventing abuse and ensuring fair access), and PIPEDA compliance (consent-based data processing, purpose limitation, and individual access rights).

Fintech clients: 42 in 12 months (40% above target), totalling $2.8M CAD ARR. Client types: personal finance apps (12), lenders (9), financial advisors (8), accounting/bookkeeping (7), and other fintech (6). Bank connectivity: 99.4% uptime across all 25 financial institutions — the resilient scraping architecture reduced outages from bi-weekly to quarterly. When outages occur, 85% self-heal within 2 hours. Data quality: transaction categorisation accuracy: 93% (from 68% with raw scraped data — the ML enrichment model was the primary driver). Merchant name accuracy: 96%. Fintech clients report that the data quality enables use cases that were impossible with raw bank data. Consent management: 340,000 consumers have granted consent through the platform. Consent revocation rate: 4% (low — indicating consumers are comfortable with the consent framework). The consent management infrastructure positions the company for compliance when open banking regulations formalise. Developer experience: average time from signup to first API call: 1.8 hours. Developer NPS: +72 (measured quarterly). The developer experience was cited by 34 of 42 clients as the primary reason for choosing the platform over competitors. API volume: 8.2 million API calls daily across all clients — with p99 latency of 180ms. The platform processes 15 million transactions daily for data enrichment. Series B: the 42 clients, $2.8M ARR, and market-leading bank coverage contributed to a $45 million CAD Series B. The lead investor's thesis: "The infrastructure layer for Canadian open banking — positioned to become the Plaid of Canada." Development investment: $420,000 CAD ($315,000 USD) over 16 weeks.