Case Studies
Anonymized examples of agentic systems in action—showing problem, approach, and measurable outcomes without revealing client identities.
From Ticket Queue to Agentic Triage: AI-Assisted Support for a Growing Service Desk
A service-desk operation was scaling headcount linearly with ticket volume. We designed and deployed an agentic triage layer—LLM-driven classification, knowledge retrieval, and draft resolution with human sign-off—built around the client's existing tooling rather than a rip-and-replace.
Hardening Before the Audit: A Pragmatic Security Uplift for Internet-Facing Infrastructure
An organisation with internet-facing services had grown faster than its security posture. We ran a structured hardening program: baseline the estate, close the highest-risk gaps first, then encode the controls as automation so they enforce themselves rather than decay.
Zero-Drama Deployments: Automating Release, Rollback, and Maintenance Windows for an E-Commerce Operation
A specialist online retailer depended on manual deployments and ad-hoc maintenance pages, meaning every release carried revenue risk and required out-of-hours heroics. We built an automated release pipeline with graceful maintenance-mode handling, health checks, and one-command rollback.
Private AI Infrastructure: Building a Local LLM Stack for Secure Agent Development
A private local AI stack gave agent development a controlled place to test models, observe behaviour, and reduce reliance on external inference for day-to-day engineering work.
NetBox-Driven Infrastructure as Code: Making the Inventory the Source of Truth
An IT operation managing 100+ devices and services relied on manual inventory and drift-prone infrastructure-as-code. We made NetBox the source of truth and built agent-driven reconciliation between inventory and deployed state—dry-run by default, with tiered approvals and an immutable audit trail of every agent action.
What These Case Studies Show
Each case study demonstrates a different facet of agentic and AI work: service desk automation, security hardening, release automation, private AI infrastructure, and inventory-driven infrastructure as code. The names are anonymized, but the engineering patterns and lessons are real.