Cubic: Scaling AI Agents for Mission-Critical Work

— CHALLENGE

Document-Heavy, High-Consequence Operations at Global Scale

Cubic Transportation Systems designs, integrates, and operates complex fare collection and transit systems for government agencies and commercial partners worldwide. Their operations span contracts, engineering requirements, compliance mandates, and multi-stakeholder change management - all document-heavy and high-consequence.

The company faced mounting enterprise AI adoption barriers. Thousands of high-variance documents - contracts, CDRLs, technical specifications, change orders, and RFPs - required repeatable extraction and interpretation across teams. Skilled personnel were spending disproportionate time on scope determination, requirement traceability, and downstream system updates - essential work that doesn't scale linearly with growth.

Security expectations were non-negotiable. Cubic required rigorous logging, monitoring, access controls, and a pathway to on-premises deployment for classified defense workloads. They also needed commercial cloud infrastructure that met government-grade standards.

Critically, Cubic wasn't looking for a chatbot. Their vision was to find a technology-plus-services partner that could apply AI across a wide range of use cases, move fast, and help build toward a future internal AI Center of Excellence - with enabling services eventually transitioning in-house.

The objective:

• Deploy a production-grade AI Agent platform across multiple enterprise workflows
• Deliver AI-powered contract intelligence, scope assessment, requirement extraction, and CRM automation
• Achieve measurable accuracy and enterprise-wide adoption within aggressive timelines
• Integrate AI agents with Salesforce, Jira, and Microsoft Teams
• Establish a Human-in-the-Loop governance model suitable for regulated industries
• Build toward a unified AI Agent ecosystem with a path to an internal Center of Excellence

— SOLUTION

Platform-Plus-Delivery: From PoC to Production at Sprint Speed

After completing an AI proof of concept and evaluating multiple vendors, Cubic down-selected Enjo as their strategic partner in June 2025 and moved rapidly toward contracting. The selection was driven by Enjo's platform-plus-delivery model, proven PoC performance, Center of Excellence alignment, and deployment flexibility across cloud and on-prem environments.

Dedicated infrastructure and knowledge operationalization

Troopr Labs deployed Enjo powered AI apps on dedicated secure cloud infrastructure with a 2-week sprint cadence. Enterprise content — contracts, specifications, policies, historical records - was operationalized into agent-ready knowledge structures.

Contract Intelligence Core

For contract intelligence, Enjo's AI models were trained on Cubic's contract corpus to perform interpretation, clause extraction, and downstream automation. The Contract Intelligence Core was delivered to production in under 3 months.

Rapid use case expansion

Additional use cases - Scope Assessment, Requirement Extraction, and Change Order-to-Salesforce integration - shipped in under 2 months each.

Human-in-the-Loop governance

Cubic and Troopr Labs aligned on a champion-led, Human-in-the-Loop rollout that improves accuracy through real user feedback and drives adoption through visible internal advocates. Monthly executive governance reviews track adoption, accuracy, and business impact. A comprehensive security review was completed including formal questionnaire responses, safeguards documentation, and monitoring assurance.

Expanding the AI Agent ecosystem

With foundational agents in production, the platform is now expanding into a Jira AI Agent, Product AI Agent, Universal AI Agent (a concierge-style router via Microsoft Teams), and RFP AI Agent - each leveraging shared knowledge, integrations, and governance to deliver compounding returns without compounding complexity.

— BUSINESS IMPACT

Production-Grade Accuracy Across Mission-Critical Workflows

Results

• ~92% Contract Intelligence accuracy - delivered to production in under 3 months
• >90% Requirement Extraction accuracy - thousands of requirements processed in minutes
• 32 projects live in the Scope Assessment Tool in production
• 40+ AI Agents deployed or in active development across the enterprise

In our last leadership meeting, the Head of Legal specifically called out the AI contract intelligence solution. Delivered in under three months, fully operational, and already changing how the team works.
— Alistair Burns, Chief Information Officer, Cubic Transportation Systems