How to Build an AI Usage Policy in One Afternoon
AI & Technology

How to Build an AI Usage Policy in One Afternoon

June 30, 2026

Back to Blog

You do not need a lawyer or a committee to write a functional AI policy for your care program. You need two hours, a clear framework, and the right questions. Here is the complete process.

Why Most Facilities Wait Too Long

The most common reason care facilities do not have an AI policy is that it feels like a large, complex undertaking. In reality, a functional AI policy can be written in an afternoon. It needs to answer six questions clearly, be communicated well to staff, and be reviewed annually.

Here is the complete process.

The Six Decisions Every AI Policy Must Make

Decision 1: Which AI tools are currently being used or will be introduced?

Walk through your staff's actual day. Where are they using AI - or where would they find it useful? Common entry points: activity planning, newsletter drafting, documentation support, research. Be honest about what is already happening.

Decision 2: What is your pseudonymization requirement?

This is the single most important decision in your policy. The answer should be: before describing any participant to any AI tool, staff must use a pseudonym - not the participant's real name - and must not include room numbers, dates of birth, or other identifying information. This requirement is non-negotiable from a privacy standpoint.

Decision 3: What will require administrator approval vs. what is permitted by default?

Useful default: permitted without approval - AI for activity content generation, communication drafting, and general research. Requires approval - any AI tool that claims to integrate with clinical data, any new platform not covered by the policy.

Decision 4: Who reviews AI-generated content before it is used?

The answer: any staff member who requests content is responsible for reviewing it before use. You, as director, are responsible for content that goes into official documentation or participant-facing materials.

Decision 5: What is the incident reporting process?

One clear paragraph: stop using the tool, report to [name] within [timeframe], document what was shared. The goal is that incidents get reported, not that reporting is burdensome.

Decision 6: When will the policy be reviewed?

Set an annual review date. The AI landscape is changing quickly.

The One-Afternoon Process

Hour One: Make the Six Decisions

Work through each decision above alone or with one other person. Write brief notes on your answers. By the end of the first hour you should have your list of AI tools, your pseudonymization language, your approval thresholds, your review process, and your annual review date.

Hour Two: Write and Format

Open the AI Policy Template from our companion article ("The Complete AI Policy Template for Adult Day Centers") and fill in your organization's specific information. You are not writing from scratch - you are customizing a working template.

Format the document to one to three pages. Print a draft and read it as if you are a new hire who knows nothing about AI. Does every paragraph make sense? Is the pseudonymization requirement clear? Is the incident reporting process simple enough to follow under stress?

The Review Process

Three reviews before finalizing:

Your administrator. Share the draft and ask for any organizational concerns or modifications. Usually 15 minutes is enough.

A peer. Ask another activity director or trusted colleague to flag anything confusing from a frontline perspective.

Legal counsel (optional but recommended). A 30-minute review can catch regulatory issues, particularly valuable if you serve populations with additional privacy protections (IDD, behavioral health).

Implementation: The Critical Part Most Organizations Skip

A written policy that is not communicated to staff is not a policy - it is a document.

Introduce it at a staff meeting. Walk through the key points, especially the pseudonymization requirement, and leave time for questions.

Make it available. Post a copy in the break room. Add it to onboarding materials.

Collect acknowledgment signatures. The template includes a signature line. Keep these on file as documentation that training occurred.

Reference it in supervision. When AI comes up in one-on-ones or team meetings, mention the policy.

What Good Looks Like 12 Months Later

A year from now, a well-implemented AI policy means:

  • Staff confidently use AI tools for activity content and communication drafting, following consistent privacy practices
  • When a new AI tool is introduced, there is a clear process for evaluating and approving it
  • If a staff member accidentally shares identifying information, they know exactly what to do
  • The policy has been reviewed and updated to reflect any changes in tools, regulation, or organizational practice

That outcome is achievable from an afternoon of work today.

DayGuide AI and Policy Implementation

If you are introducing DayGuide AI as part of your policy rollout, the platform simplifies implementation significantly. Because DayGuide AI's zero-PHI architecture structurally prevents participant identifying information from ever entering the system, the privacy compliance for DayGuide AI use is handled by the platform's design. Staff do not need to remember to pseudonymize within DayGuide AI - the system does not accept real names.

For general-purpose AI tools used for other tasks, your policy's pseudonymization requirement provides the guardrail. Together, the policy and the platform create a practical, sustainable approach to responsible AI use.

Ready to transform your activity program?

See how DayGuide AI saves activity directors 10+ hours per week.

Request a Free Demo