Problem
Two families of integration, both trending toward one bespoke project each.
Service scheduling data arrived from OEM partners in wildly inconsistent formats — CSV, HTML, plain text, vendor-specific metadata — for Xtime, CDK and ReyRey. Separately, service providers each needed their own integration: VUMO, HopDrive, MyKaarma, FedEx, and Axle for insurance validation.
Another team had already estimated a comparable integration at 3 months apiece. At that rate the roadmap was years of the same work repeated.
Action
Built one input-agnostic parser architecture that normalized any of those formats into a single pipeline, and a reusable service architecture underneath the provider integrations — instead of a bespoke parser and a bespoke integration per source.
Result
VUMO and HopDrive were each integrated in under 1 week, against that 3-month per-integration estimate from elsewhere in the org.
The providers that followed onboarded near-seamlessly, which was the actual point: the first integration was the only one that cost what an integration normally costs.