SaaS Development Company in Chicago

Logistics / Supply Chain

A Chicago-based logistics technology startup (Series A, $12 million raised) was building a freight visibility platform — SaaS that tracks shipments in real-time across multiple transportation modes (truck, rail, ocean, air) and provides predictive ETAs to shippers and receivers. The founding team had built a working prototype that tracked truck shipments using ELD (Electronic Logging Device) data — but the platform needed to scale from truck-only to multimodal, and from 50 beta customers to 500+ enterprise customers. The prototype's problems: single-mode only (truck via ELD data — no rail, ocean, or air tracking, which meant the platform couldn't provide end-to-end visibility for intermodal shipments that pass through Chicago's rail hubs), accuracy degradation at scale (predictive ETA accuracy was 78% within a 2-hour window for 50 customers but degraded to 62% when tested at 500-customer load — the ML model wasn't designed for multi-tenant data isolation, causing cross-customer data leakage in predictions), no enterprise features (no SSO, no RBAC, no audit logging — enterprise shippers like Walmart, Procter & Gamble, and Unilever require these features before procurement will proceed), and API-only interface (the prototype had no web UI — customers accessed data through API calls, which worked for developer-first beta users but not for logistics coordinators who need dashboards and alerts). The startup needed to close 5 enterprise deals ($1.5M combined ARR) within 9 months to raise a Series B.

We rebuilt the platform as a multimodal freight visibility SaaS with enterprise features and production-grade predictive accuracy. (1) Multimodal data integration: expanded from truck-only to four transportation modes. Truck: ELD data via KeepTruckin (now Motive), Samsara, and Omnitracs APIs — GPS location every 5 minutes, plus HOS (Hours of Service) data for driver availability predictions. Rail: integration with Class I railroad APIs (Union Pacific, BNSF, CSX, Norfolk Southern) for train tracking, plus RFID-based container tracking at intermodal terminals — critical because Chicago is where most east-west and north-south rail traffic converges. Ocean: vessel tracking via AIS (Automatic Identification System) data aggregated from MarineTraffic and VesselFinder, combined with port terminal operating system APIs for container yard status and gate-out events. Air: integration with airline cargo tracking APIs (United Cargo, American Airlines Cargo, FedEx, UPS) and airport cargo handling system data. Data normalisation: each mode produces data in different formats, frequencies, and coordinate systems — we built a normalisation layer that converts all tracking data into a unified event stream (timestamp, location, velocity, status, mode, carrier, shipment reference). (2) Predictive ETA engine: replaced the single logistic regression with an ensemble model. Historical patterns: trained on 18 months of shipment data — 4.2 million completed truck deliveries, 800,000 rail moves, 200,000 ocean legs. The model learns route-specific patterns: the I-80 corridor through Indiana slows 15-20 minutes during Chicago rush hour, rail terminal dwell at Chicago's intermodal yards averages 14 hours but varies by terminal (BNSF Logistics Park is faster than Norfolk Southern Landers), and ocean port congestion at Los Angeles causes 2-3 day delays that propagate to Chicago rail connections. Real-time adjustments: the model incorporates live data — weather (NOAA API), traffic (HERE/TomTom), port congestion (MarineTraffic), and rail network status — to adjust predictions in real-time. Multi-tenant isolation: each customer's shipment data is isolated in predictions — Customer A's unusual routing patterns don't contaminate Customer B's ETA predictions. We achieve this through per-customer feature engineering: the model uses route-specific features rather than customer-aggregated features. Prediction output: ETA with confidence interval (e.g., "arriving 2:30 PM ± 45 minutes with 85% confidence") — logistics coordinators need to know not just the predicted time but how certain the prediction is. (3) Enterprise web platform: designed for logistics coordinators (non-technical users who manage hundreds of shipments daily). Shipment map: real-time map showing all active shipments with colour-coded status (on-time green, at-risk yellow, late red). Filtering by mode, carrier, origin, destination, and customer reference. Exception dashboard: proactive alerts for shipments that are late, at-risk, or experiencing exceptions (carrier delay, weather event, port congestion). Each alert includes: the current ETA impact, the root cause, and suggested mitigation actions (contact carrier, notify receiver, arrange alternative transport). Analytics: historical performance analytics — on-time delivery rates by lane, carrier performance scorecards, average dwell times by facility, and predictive accuracy metrics. Reports exportable as PDF and Excel for quarterly business reviews. (4) Enterprise features: SSO (SAML 2.0 and OIDC — required by every enterprise prospect), RBAC (role-based access: admin, manager, coordinator, viewer — with customer-configurable permissions), audit logging (every data access and action logged with user, timestamp, and IP — SOC 2 requirement), API with rate limiting (RESTful API with OAuth 2.0 authentication, rate limiting per customer, and webhook support for real-time event notifications), and white-labelling (enterprise customers can embed the tracking interface in their own customer portals with their branding). (5) SOC 2 Type II: implemented from project inception. The platform processes sensitive supply chain data (shipment contents, routing, pricing) — SOC 2 provides the trust framework enterprise customers require.

Multimodal coverage: the platform now tracks shipments across truck (85% of volume), rail (8%), ocean (5%), and air (2%) — providing end-to-end visibility for intermodal shipments that pass through Chicago's rail hubs. ETA accuracy: 91% within a 2-hour window (from 62% at scale with the prototype — 47% improvement). The ensemble model and real-time adjustments were the primary drivers. For truck-only shipments, accuracy reaches 94%. Enterprise deal velocity: 8 enterprise deals closed in 8 months (exceeding the 5-deal target) totalling $2.8M ARR. Average deal: $350K ARR. Average sales cycle: 10 weeks. The SOC 2 report and enterprise features (SSO, RBAC, audit logging) were cited by 6 of 8 customers as "required for procurement to proceed." User adoption: 1,400 daily active users across enterprise customers (logistics coordinators and managers). Average session: 34 minutes. Exception dashboard is the most-used feature — coordinators start their day reviewing exceptions rather than manually checking shipment status. Alert effectiveness: the platform generates 3,200 proactive alerts daily. 78% of alerts result in coordinator action (contacting carrier, notifying receiver, escalating internally). Customer NPS: +62 (measured at 6 months). API usage: enterprise customers make 2.4 million API calls daily — integrating visibility data into their own TMS, ERP, and customer portals. The webhook system delivers 180,000 real-time notifications daily. Series B: the 8 enterprise deals and $2.8M ARR contributed to a $42 million Series B at a $280 million valuation. The lead investor noted "Chicago's logistics DNA" and "institutional-grade enterprise execution." Development investment: $420,000 over 16 weeks.