Branch New Business AutomationWeb app, OCR model, Applied Epic API SDK
Brokers typed DASH report data into the BMS by hand.
A model reads the report, the broker checks it, and code writes it into the BMS.
10 to 15 minsaved per transaction
$6.2Mprojected annual value$4.3M from live solutions
28AI solutions22 live, 6 in final rollout
97%Copilot adoptionacross the initial 240+ licenses
200+at Tech Days 2026sessions I ran on AI and agents
How these are measured. Process Excellence times each process before and after we automate it and applies the department's salary rates. Revenue projects are measured on sales. I built the intake and ROI tracking behind every number here.
Selected work
Comparative Plan Analysis AgentCopilot Studio, OCR, AI Builder
Reps spent up to an hour building each plan comparison.
The agent reads the policy booklet and fills in the whole comparison.
45 secdown from 45 to 60 min
By hand45:00
Agent0:45 done
45 minutes compressed into 6 seconds
AI prospecting engineAzure Functions, AI models, CRM API
Reps researched and keyed in their own leads.
A rep picks an industry and region; scored leads land in the CRM.
30,667prospects loaded
Meridian and a modernized mobile app.NET 10, Azure SQL, Service Bus, API Management, React Native. Built by me with AI coding agents.
Leads, the CRM, sales tracking, and policies lived in four separate systems.
One CRM layer over Applied Epic, DASH, and Verisk, with a customer app feeding it. Old systems can be retired later by configuration.
Adoptedthe business is moving forward with the concept
Propensity scoringAzure ML, Azure AI Foundry
Ranks inbound prospects by how likely they are to buy.
In progress
Also live: an IT service desk triage agent, a SharePoint knowledge assistant, and a SQL monitoring agent.
The team
One developer in July 2025, two more in October, four more in May 2026. Seven developers and a business analyst today.
- Intake
Requests come through the service desk, ranked by hours and revenue.
- Learn
A developer works with the team directly and learns the job as it's really done.
- Build
Mostly code and direct API integrations, with agents where they fit.
- Measure
Before and after numbers, reported to executives every month.
How I build
I build with AI coding agents. They write the code. I describe exactly what's needed, set the method and acceptance criteria, and know the infrastructure it runs on.
- AI
- Azure AI Foundry, Azure ML, Azure AI Search, RAG
- Cloud
- .NET, Azure Functions, Service Bus, API Management, Azure DevOps
- Microsoft 365
- Copilot, Copilot Studio, Power Automate, AI Builder
- Governance
- Evals and LLMOps, security reviews, data kept in Canada, IS risk audits passed
What I've learned
- Done means adopted
- Measure ROI and completion by business adoption. Finishing the build is only half the battle. Training, deployment, and continuous improvement all come after it.
- Be willing to stop
- A deeper look often brings out the truth, even when it isn't what you were hoping for. It takes a strong leader to say plainly that this isn't the best path forward. Too many projects keep going just because they were approved. If the return isn't there, sound the alarm, with the evidence to back it up.
- Adoption is a people play
- AI adoption isn't a technology play, it's a people play. We're creatures of habit, so it takes guidance and change management more than an IT rollout plan. Help people build habits, start with the mundane tasks, and let them get comfortable. Adoption follows.
- Clear requirements win
- Building with AI agents has cut our prototype and development time, and that gap will keep growing as models improve. The winners will be the people who can say clearly what they need and how it should be built. Even today, an agent produces working code when it gets proper requirements and architecture.
- Let the team fly
- Growing a team means letting go of some control. People thrive when they get the runway and the responsibility. Real growth comes through trial and error, and through wins earned in tough moments. Give guidance, let go of the reins, and see where they take you.
15 years, service desk to AI
- 2011Support analyst, sysadmin100+ offices
- 2012Team Lead, Tech SupportRan the service desk
- 2015Manager, App SupportHUB International
- 2017Systems Administrator~600 servers
- 2019Enterprise AdministratorFreshservice to 2,000+ users in a month
- 2025AI Innovation & SolutionsStarted the function; Manager since 2026
Western Titans 2026The company's highest honour
Western Champion 2014Top ~80 of 1,800 staff
Microsoft CertifiedAzure AI Fundamentals