Employee Relations Case Intake Framework

Go to claude.ai, open Projects, and start a new one focused specifically on the triage step, not on drafting investigation findings.

1. Click Projects in the sidebar, then New project
2. Name it something like "ER Case Intake Framework." Keep this separate from any project you might use for drafting an actual investigation's findings
3. Fill in the project description: this project helps frame a new ER complaint, categorize it, and identify next steps, before an investigation begins

What you should see: A new, empty project ready for instructions and knowledge files.

Before uploading anything, describe what a good intake framing looks like at your company.

1. List your standard complaint categories (conduct, performance-adjacent, conflict between employees, policy violation, safety, other)
2. State what every framing should include: category, relevant policy section, initial documentation checklist, and a flag for whether outside counsel involvement should be considered
3. Save the instructions

What you should see: Instructions saved and visible in every new chat you start inside this project.

Add your written ER policy and intake checklist to the project's knowledge panel so every framing references the same source material.

1. Open the Files/Knowledge panel
2. Upload your ER policy or complaint-handling procedure (PDF, Word, or plain text)
3. Upload your intake checklist if you have one separately

What you should see: Both documents listed in the project's knowledge panel, available to reference in every chat from now on.

When a new complaint comes in, open a chat inside this project and describe the situation in general terms only.

1. Describe the type of complaint, the working relationship involved (peer, manager-report), and roughly how long the issue has been going on
2. Do not include the complainant's name, the accused's name, or any specific identifying project or client detail
3. Ask for a framing: likely category, relevant policy section, and an initial documentation checklist

What you should see: A structured response identifying the likely complaint category, pointing to the relevant section of your uploaded policy, and listing the first few documentation steps to take.

Treat the "should counsel be looped in" flag as a prompt to make that call yourself, not as the final answer.

1. Review the framing's flag on counsel involvement
2. If the situation touches conduct that could carry legal exposure (harassment, discrimination, retaliation, safety), raise it with your legal partner before doing any further documentation, regardless of what the framing suggested
3. If counsel decides to direct the investigation, stop using a general AI tool for any further work on that case. Counsel may need the entire process to run through their own systems to preserve attorney-client privilege, and drafting outside of that structure can risk waiving it

What you should see: A clear decision, made by you and your legal partner, about who owns the next steps on this case, documented outside of this project once that decision is made.

If the complaint touches an accommodation, a medical leave, or any health-related detail, exclude that detail from the framing conversation completely, even in general terms.

1. Describe the complaint by its behavioral or policy category only ("a conduct complaint involving repeated conflict"), never by any medical or diagnostic detail
2. If the case is really about an ADA accommodation or FMLA leave dispute rather than a conduct issue, route it to your leave and accommodations process instead of this ER intake framework. Those categories have their own confidentiality requirements that a general framing tool shouldn't touch
3. Check your obligations under the ADA and FMLA, both of which require medical information to be kept confidential and stored separately from the personnel file, with your legal or compliance team if you're unsure which process a case belongs in

What you should see: An intake record that documents the behavioral complaint clearly, with zero medical or health information anywhere in the conversation.