Describe the work
What takes too much time, judgment, follow-up, or rework? What would better look like?
Ten Champions. Real work. Three frames for moving from what AI can help one person do to what Our House can responsibly build, share, measure, and govern.
Make the work visible. Practice prompting, skills, assistants, agents, and workflows.
Find the patterns that can travel across roles, houses, and operating teams.
Choose the policy, platform, pilot, ownership, and evidence pathway.
Monday is not a model bake-off. It is a practical working session about how each Champion can make one recurring workflow clearer, safer, and more repeatable.
What takes too much time, judgment, follow-up, or rework? What would better look like?
Prompting for a better answer. A skill for repeatability. An assistant for context. An agent for bounded recurring work.
What data may be used? What must be reviewed? What requires IT, compliance, legal, or leadership verification?
Time saved, errors reduced, follow-up improved, training made more consistent, or a decision made more clearly.
“Show us the work you actually do. Where does it slow down, repeat, or require more attention than it should?”
These are not promises of automation. They are working hypotheses captured in the advance Champion sessions and ready for Monday’s conversation.
Prompt and skill patterns, Markdown as portable memory, HTML-first reporting, and bounded marketing prototypes.
Leadership progress, audit dashboards, SharePoint retrieval, and decision support for regional operations.
Job-posting review, bus-trip calendar maintenance, and concise occupancy-report consolidation.
Non-PHI walkthrough, infection-control, and med-cart audit support are possible exploration lanes.
Executive sponsorship, platform standards, shared skills, governance, and a 30/60/90-day path.
Compliance writing and investigation support can become a safe-use lane when source, access, and review are explicit.
Care-level estimator, sales and marketing guide, room checklist, tour practice, and website chatbot concepts.
MedCart comparison, assessment follow-up, missing care hours, and a review queue from approved or sanitized data.
Track required approvals, draft follow-ups, summarize regional exceptions, and coach leaders with visual SOPs.
Rent-roster retrieval, invoice reconciliation, care-level variance, Sage Intacct exports, and usage analysis.
HIPAA is spelled HIPAA. A tool is not “HIPAA compliant” in the abstract. Our House needs to verify a specific plan, tenant, configuration, data flow, workflow, and oversight model.
What is allowed, restricted, prohibited, and subject to approval?
Which commercial plan, tenant, account, region, and admin controls are in use?
What may enter the system? What is PHI, resident, employee, financial, or public data?
May AI draft, summarize, recommend, retrieve, or execute? What is the minimum necessary access?
Who reviews, logs, audits, disables, and owns the outcome?
Public information, generic training, brainstorming, de-identified examples, and administrative work without sensitive data.
Internal documents, employee information, operational reports, finance exports, resident-adjacent work, connectors, and shared assistants.
PHI, clinical records, medication data, resident identifiers, autonomous chart changes, corrective action, or automatic external communication.
“What exact configuration would let us test one narrow workflow with the minimum necessary data, a signed BAA where required, role-based access, auditability, and human approval?”
What can I do better? What can I preserve as a skill? What should remain human-reviewed?
Leave with: one workflow and one safe experiment.What patterns travel across roles? What can be shared? Where do permissions and context diverge?
Leave with: team patterns and ownership questions.What should Our House allow, govern, measure, pilot, and fund?
Leave with: decisions, pilots, and a 90-day roadmap.Begin with what is possible. Make safety part of the design. Measure the value that earns the next step.