
A technical inventory for Crane Worldwide Logistics IT leadership. Enough detail to evaluate the platform. A deeper due-diligence packet is available under NDA on request.
WakeTech.ai runs on Microsoft Azure. Every enterprise customer receives a dedicated deployment consisting of their own compute, their own storage, and their own database. No shared infrastructure, no shared schema, no noisy neighbor. For customers with strict data residency or compliance requirements, deployment can land inside the customer's corporate Azure tenant under their own billing and governance.
The deployment automation is infrastructure-as-code, so spinning up a new enterprise customer environment is a repeatable and audited process, not a manual click-through.
The core platform is written in TypeScript on Node.js, served through Next.js for the web application layer. TypeScript enforces type safety across the codebase, which materially reduces the class of bugs that reach production. Next.js gives us server-side rendering for fast initial page loads, API route handlers for integration endpoints, and a static build pipeline that deploys as a single standalone artifact.
Transactional data lives in Microsoft SQL Server on Azure. Every customer has a dedicated database instance. No shared database. No logical tenant separation inside a common schema. SQL Server was chosen because it is the industry standard for freight and logistics accounting workloads, integrates cleanly with Microsoft Azure, and has a mature backup and point-in-time recovery story that enterprise procurement teams are already familiar with.
Object storage for documents, BOL images, EDI archives, and other binary artifacts uses Azure Blob Storage with per-customer containers and server-side encryption. Retention policies and lifecycle rules are set per customer according to their compliance requirements.
The WakeTech.ai AI crew runs on Anthropic Claude models via the Anthropic API. Claude was selected for three reasons: strong enterprise security posture (SOC 2 Type II, no training on customer data, zero data retention available), high reasoning quality for operational decision-making tasks, and a mature tool-use API that supports the agent architecture the crew is built on.
Agents are stateful services with their own memory stores, skill catalogs, and approval thresholds. Each agent (Lane, Ada, Pulse, Scout, Ledger, Intake, Sentinel, Apollo, Rune, Vera) is scoped to a specific operational domain and earns autonomous authority through demonstrated performance against an agreement-rate threshold set per workflow.
Corridor intelligence (WakeSignal) is a separate analytical layer scoring US freight corridors against NOAA weather data, EIA energy signals, USDA agricultural signals, AIS vessel tracking, and a proprietary hurricane forecast model. This runs on our own infrastructure, not through a third party, and does not ship customer data to external analytics services.
Truck routing is powered by a self-hosted Valhalla routing engine running against OpenStreetMap data for North America. Valhalla is a respected open-source routing project used by Mapbox, government agencies, and commercial logistics operators. Self-hosting means WakeTech.ai customers pay zero per-request fees for routing, zero per-seat licensing, and get sub-second route calculation with full truck profile support (vehicle dimensions, weight, hazmat, bridge restrictions, time-of-day access rules).
Map rendering uses PMTiles with MapLibre GL for vector tile delivery. PMTiles is a single-file archive format that serves vector tiles over HTTP range requests without a tile server dependency. MapLibre is the open-source fork of Mapbox GL and does not require a Mapbox access token or per-view billing.
EDI is handled natively by WakeEDI, our production EDI processing service. It currently handles the core freight transaction set: 204 tenders, 990 tender responses, 214 shipment status updates, and 997 functional acknowledgments. WakeEDI supports SFTP and AS2 transport, accommodates partner-specific quirks without custom development per partner, and logs every message for audit and replay.
General-purpose integration uses a combination of REST APIs exposed by core, webhook event subscriptions for async notification, and customer-specific adapters that live in the customization layer. WakeTech.ai integrates with 94 distinct external services today, including DAT, Highway, Greatwide, FMCSA, NHTSA, NOAA, Google Places, ElevenLabs, Twilio, TomTom, Mapbox, OpenDock, Ryder, and TruckerTools.
WakeTech.ai operates its own self-hosted JMAP and IMAP mail infrastructure for agent mailboxes, dispatch communication, and transactional customer email. Running mail on our own infrastructure keeps sensitive freight communication out of third-party mail providers, avoids per-mailbox licensing costs, and gives each customer the option of a dedicated mail domain for their deployment.
Outbound notification supports SMS and voice through an A2P 10DLC compliant messaging pipeline for driver and dispatch alerts. Voice agents for live carrier negotiation use enterprise-grade real-time voice synthesis.
Every deployment runs its own monitoring, logging, and alerting stack. Infrastructure metrics, application logs, and business event streams are collected per customer and retained according to customer policy. The WakeTech.ai Sentinel agent provides automated infrastructure health monitoring and can escalate incidents before they become customer-visible outages. Customers have direct read access to their own operational telemetry.
Each customer deployment is a separate network perimeter. There is no cross-tenant query path at the infrastructure level, only at the application level, which itself is scoped to a single customer database.
TLS 1.2 or higher for all traffic in transit. Encryption at rest for every database and every storage container. Customer-managed keys available on request.
Application service accounts have only the database permissions they need. Schema changes require elevated credentials held by a separate operations account. Secrets are stored in Azure Key Vault.
Every material action in the platform is logged with actor, timestamp, and affected entity. Audit logs are append-only from the application perspective and tamper-evident through infrastructure controls.
A due-diligence packet covering vulnerability response policy, penetration test history, dependency management, and supply chain controls is available under NDA on request.
Enterprise IT teams evaluating WakeTech.ai often ask about things we explicitly do not use.
Every customer has their own database instance. We do not mix customer data in a shared warehouse with logical tenant filtering. This is a deliberate rejection of the dominant cloud-native SaaS model. The trade-off is a slightly higher per-customer infrastructure footprint in exchange for infrastructure-level data isolation that application bugs cannot compromise.
We do not call Google Maps, Mapbox Directions, or PC*Miler for routing or map rendering in production. All map and routing capability is self-hosted. Predictable cost, no per-seat licensing, no per-query bill that scales with customer revenue.
We do not ship customer operational data to general-purpose analytics or BI tools outside the customer deployment. Analytics runs inside the dedicated environment and does not leave it.
Anthropic has committed in their enterprise terms that customer data sent through the API is not used to train Claude. WakeTech.ai does not train its own models on customer data either.
Customer-specific workflows and extensions live in a dedicated customization layer scoped to that customer deployment, not merged into the shared core that other customers run.
Exact version inventories, CVE response policy, penetration test reports, SOC 2 posture, dependency tree, deployment runbooks, and disaster recovery documentation. Questions about any component on this page go directly to the person who built it.