Services and approach
We work through one process from the first request to the finished task. I help you decide what needs to change and where AI is useful.
We trace one process with the people who run it, including the delays and workarounds. The recommendation may be to fix the manual process before adding AI.
What you receive: A process map and a short list of changes to make first, including whether AI would help.
We work out what information people need, who can approve a change, and who carries it out. If Workday is involved, we connect those decisions to the business process and reporting. We agree on how to check whether the change helps.
What you receive: A new process map, a review template, approval rules, and tests for likely mistakes.
I look at the answers the pilot produced, the checking and correction they needed, and the total time involved. We test whether people can stop a mistake and whether the tool stayed within its agreed use.
What you receive: A written review of the results, required fixes, and whether to expand, change, or stop.
We agree on scope, timing, and a fixed project fee before starting. These time ranges are a guide, not a commitment before we have discussed the work.
Bring an organization change, a performance issue, or a difficult conversation you need to work through. We can talk through the situation and prepare for what you need to do next. No AI tool required.
We’ll agree on the focus, session length, and fee before scheduling.
Ask about HRBP coaching$150/hour
We can review your AI setup or work through a task in Claude or ChatGPT together. That might mean improving the context you give it, checking an answer, or saving work you want to reuse.
One session is fine. Email me with what you use now and what you want help with. We’ll agree on the length before scheduling.
Ask about a sessionMade-up example · No employee data
A manager attaches an old travel-and-expense policy. The shared library has a newer version. The immediate task is to work out which one applies. The process also needs a way to prevent the old link from causing the same problem next week.
Travel and expenses
Later effective dateConfirm the expense date, employee group, and any approved exception.
The policy owner confirms what applies. Nothing is sent automatically.
Human review required
My recommendation: Fix intake and source ownership first. The tool should stop if the approved policy is missing, versions conflict, or the question needs legal interpretation.
This is a sample of a deliverable, not a client case study or a deployed workflow.
Use the printable workflow worksheetI test what happens when a source changes, information is missing, or a step fails. A reviewer should be able to spot the problem and stop the next action. Their corrections need to remain attached when the process runs again.
Use approved material. Keep its source, version, date, and scope with the work.
Route the request by issue, location, sensitivity, and owner.
Prepare a brief that separates facts from interpretation and labels assumptions.
Look for conflicting sources, missing facts, and reasons to stop or ask for help.
A named person can correct, reject, approve, or escalate the proposal. They need time and authority to do that.
Take only the approved action, with no more access than the task needs.
Check that the action happened. Make failures and exceptions visible to the owner.
Save the sources, open questions, reviewer’s corrections, decision, and evidence of completion together. A later run must preserve the corrections.
That is what I mean by auditable: someone can check what information was used, who made the decision, and whether the approved action happened.
What to keep in the recordI work on approved-document review, policy-question intake, HRBP briefs, and the handoffs that follow an approval. AI can help gather information and prepare a draft for review.
I do not build tools that make employment decisions or give legal advice. Decisions about discipline, termination, pay, promotion, accommodations, leave, or investigations stay with qualified people.
Work involving real company data or production systems needs the client’s technology, security, privacy, legal, and business owners. A pilot review is not a compliance certification or a security assessment.
For the first conversation, leave out passwords, employee records, and sensitive company material. We will agree on how to handle anything needed later.
When I would leave AI outHR Mission Control is my private prototype, tested with public and made-up data. Its case study includes dated checks, a duplicate-case failure, and what I changed. Those results are not client outcomes or proof of production readiness.
Read the HR Mission Control case study