Complete CRM/ERP platform for a German insurance & social-care company. Replaced ~230 Make.com scenarios, 26 JotForm forms and several Monday.com boards.
Challenge
The client had a tangled automation zoo: ~230 Make.com scenarios (104 active), 26 JotForm forms, 4 Monday boards, SuperChat templates for WhatsApp/SMS/Email. The system worked but was opaque — it broke under load, made testing impossible, and had no end-to-end analytics or audit trail. Workers were assigned manually via spreadsheets; time and payroll tracked across multiple systems.
What We Did
- Microservice architecture on .NET 8 with MassTransit bus, CorrelationId tracing, distributed logs in Dash0 via OpenTelemetry.
- 16 microservices: Core.BusinessLogicService (41+ DbSets, 22 controllers), CallCenterService (Cisco Webex), ActivityHistoryService (17 event consumers), DocumentGenerationService (9 template types), MondayIntegrationService, LearningService (LMS + proctoring), ManagerTimeTrackerService (time tracking + DATEV).
- 130+ permission flags across domains: clients, contacts, call queues, workers, insurers, partners, complaints, reports, time tracking, tasks, documents, social services and Learning.
- Two-stage distribution: PostGIS radius filter + Google Maps Routes API for precise arrival time. Scoring across 8 criteria (max 23 pts). 15 EventHandlers recalculate in cascade.
- NEW v9.0: ManagerTimeTrackerService — 8 controllers, 5 DbSets, 3 background services. One-click DATEV export.
- Bidirectional GraphQL+webhook sync with Monday. PersonMatcher with 3-tier logic (email → name+birthdate → name).
- In-house call centre on Cisco Webex with queue routing CC1–CC4 and Recall logic (3 attempts, callback windows 10–13 and 15–17).
Result
- Migrated from legacy Make/JotForm/Monday stack to a proprietary platform.
- End-to-end analytics and audit via ActivityHistoryService with CorrelationId and change snapshots.
- MP4 fix (Apr 2026): resolved status drift for 31 of 104 workers in production.
- Active development: Apr–May 2026 — 47 commits in Core.BusinessLogicService, 80 in Frontend.PortalService.
- Production metrics: 8 regional DieAlltagsfeen sites monitored every 60 s, avg response 286 ms, status page covers 42 monitors and 3 domains.