MY CONTRIBUTION
- Sessions
- 34
- Registrations
- 1,412
- Attendees
- 1,019
- Avg. satisfaction
- 2.91 / 3
Participation and satisfaction for the whole series.
HR & BENEFITS TECHNOLOGY | AI ENABLEMENT | PROCESS IMPROVEMENT
Helping teams improve benefits processes and the systems behind them.
I worked in member support and benefits for more than ten years before moving into IT project delivery in December 2024. I help teams fix the workflows and systems employees depend on. I test the changes, write the guidance, and use automation and AI where they fit the work.
- Location
- Plymouth, MI
- Open to
- HR and benefits technology, AI enablement, process improvement
- wshelton89@gmail.com
Selected Work
Benefits work showed me where systems break down for employees, like rejected eligibility files and data that does not match between vendors. I use that when I gather requirements, test a workflow, or write the guidance people will read.
View work for
Portfolio examples
Independent projects that show how I would approach a problem, separate from the workplace work above.
-
Portfolio simulation
AI Opportunity Intake Workbench
When a team has more AI ideas than time, it has to decide which ones to pursue first. This simulation shows how I would collect, score, and route those ideas for review. It is documentation and templates.
Repository private while I revise it. Available on request.
-
Portfolio governance example
Policy Copilot Governance Demo
A written plan for a narrow HR policy assistant in Copilot Studio: what it may answer, when it hands off to a person, how to test it, and what to measure. It is documentation only, with no live bot to try.
Repository private while I revise it. Available on request.
-
Portfolio process-design example
Offboarding Process Redesign Case Pack
A sample offboarding redesign: a problem statement, SIPOC, user stories, RAID log, KPI framework, and process notes. It is separate from my workplace Power Automate flow.
Repository private while I revise it. Available on request.
-
Portfolio framework
AI Readiness Assessment
A documentation-first framework for deciding whether a team is ready to adopt AI. It scores six dimensions and ends with a go, fix-then-go, or no-go recommendation. One example is process maturity: is the workflow written down well enough that someone could automate part of it without re-interviewing everyone? It has not yet been run with a real team.
Repository private while I revise it. Available on request.
Perplexity and Computer Projects
I use Perplexity and Perplexity Computer extensively for research and project work. I chose the problems, set the requirements, and reviewed and refined the outputs, with AI assistance. These are personal projects, separate from any employer.
- Reusable skillsConfigured Computer skills with defined inputs, task ownership, output formats, and approval steps
- Multi-step workflowsResearch, evidence review, and decision-ready handoffs instead of one-off prompts
- Human approvalConsequential actions wait for my review; plans and results are kept separate
- A published learning toolAn interactive simulator anyone can try in a browser
-
Independent project
Reusable Perplexity Computer Skills
I set up and refined a library of Computer skills for research, evidence management, communication, and personal workflow automation. When I write a skill, I spell out what it needs as input, what it is responsible for, what the output should look like, and where it has to stop and ask me before acting. When one misses, I adjust it.
-
Independent project
Evidence Controls and Perplexity-to-Claude Handoffs
I use Perplexity and Claude for different parts of my research and project work. I set rules for which records each tool reads first, what each one is allowed to change, and which claims are restricted. Work passes between them as dated handoff notes. I review a handoff before I rely on it for a decision or approve an action that reaches anyone else.
-
Published learning tool
Offside Lab
An interactive simulator I built and iterated with Perplexity Computer. It teaches a complex rule through draggable scenarios, freeze-frame explanations, and plain-language feedback. It shows how I turn a hard concept into something people can try for themselves.
Try Offside Lab
Built with AI assistance: Perplexity Computer generated much of the implementation.
Experience
I started in health-plan member support, then spent years in benefits: enrollment and life-event changes, COBRA and dependent eligibility, EDI 834 eligibility files, vendor data, audits, and open-enrollment presentations to groups of 300 or more. The systems were PeopleSoft, Workday, and ADP Workforce Now: I managed benefits data in PeopleSoft, corrected enrollment records and tested benefits releases with written checklists in Workday, and set up annual enrollment and configured deduction codes and eligibility rules in ADP Workforce Now.
In December 2024 I moved into IT as an IT Project Manager / AI Enablement Specialist. I handled application intake, built and piloted automation, and supported people learning new tools. The systems work started earlier, in benefits: I managed the project that replaced a paper tuition-reimbursement form with a SharePoint approval workflow, working with a colleague who brought the SharePoint and Power Automate expertise. Those years are why I ask how a change will land for the employee on the other end.
- 01Benefits operationsEnrollment support, data audits, vendor coordination, employee education
- 02Systems and IT project deliveryRequirements, Microsoft 365, Power Automate, process documentation
- 03Enablement and adoptionTraining content, user guidance, communications, user support
How I Work
Start with the process.
Understand the work and responsibilities before choosing a tool. On HR offboarding, I gathered HR and IT requirements and wrote the charter and test plan before building anything.
Choose a practical solution.
Use AI where it fits and a straightforward workflow where it does not.
Make the handoff clear.
Document the process, testing, user guidance, and responsibilities.
Separate plans from results.
State what was built, what was tested, and what remains a proposal. That is why the SCORM project here is labeled as a pilot.