These projects show how I approach workflow design beyond the interface — combining conversational AI, automation, data structure, security and governance into solutions that are designed to work in real organisations.
Each case study includes implementation detail, architecture thinking, and proof of delivery through videos, diagrams and supporting documentation.
2nd Place, Operative Badge
Client requests arrive as unstructured briefs, documents and emails — making them slow to triage, hard to risk-assess, and difficult to route and hand off consistently.
I built Engagement Hub, a multi-agent intake and delivery-prep system using Copilot Studio, Power Automate and Dataverse. It analyses client submissions, creates duplicate-safe Engagement Requests, generates discovery briefs and Word handoff documents, and notifies Teams when human review is needed — logging every step for auditability.
Multi-agent architecture: parent orchestrator, specialist Contract Analysis Agent and Delivery Prep child agent
AI document analysis extracting risks, gaps, a confidence score and a human-review recommendation
Duplicate-safe Engagement Request creation using Dataverse relationships and business-context checks
Discovery brief handoff generated and saved back to Dataverse
Word handoff documents from Dataverse-grounded output, Word templates and SharePoint storage
Idempotent Teams human-review alerts, logged in an Automation Log to prevent duplicates
Manual access requests are slow to process, difficult to track, and often lack the governance needed for secure approvals.
I built an end-to-end access request and approval solution using Copilot Studio, Power Automate and Dataverse. It manages the full lifecycle from request to approval to fulfilment, with AI summaries, SLA reminders and governance controls designed for enterprise use.
End-to-end request to approval to fulfilment workflow
Reusable AI summaries configured through environment variables
Automated SLA reminder watcher for overdue approvals
Governance controls including roles, auditing and idempotency
IT support workflows become unreliable when ticket ownership, status access, and escalation decisions depend on user-entered information rather than verified identity.
I built an authenticated IT support agent for ticket logging, status checks, urgent escalation, and AI-powered ticket summarisation using Copilot Studio, Power Automate, Dataverse, and Azure OpenAI. The solution captures trusted Entra ID user identity, uses those values to authorise ticket access and escalation, and separates notification handling into a watcher flow to improve reliability and avoid duplicate alerts.
Trusted ticket access using signed-in Entra ID identity
AI-generated ticket summaries for cleaner operational context
Identity-based status checks and escalation authorisation
Watcher-based alerting with duplicate-notification protection
Technician console for operational triage and follow-up
I design solutions around process flow, data structure and operational reliability — not just the front-end experience.
I build with identity, roles, auditability and controlled update paths in mind from the start.
My work focuses on real internal workflows such as approvals, escalations, lifecycle tracking and exception handling.
Each project is documented through walkthrough videos, diagrams and written detail so both technical and non-technical audiences can understand the solution.
Each project page includes a deeper breakdown of the problem, architecture, implementation decisions and governance considerations, along with video walkthroughs and supporting assets.