Medication lists disagree
Nine meds, two instructions, one family left to spot the conflict.
HomeRelay turns discharge instructions into a safe, shared first 72 hours at home.

The hard part is turning a packet into safe work that is timed, owned, and clear enough for a tired family to trust.
Nine meds, two instructions, one family left to spot the conflict.
The packet says cardiology within seven days. Nobody has booked it.
The family needs the few symptoms that mean call now, not pages later.
Pickup, rides, setup, and check-ins drift until someone claims them.
He should not have to coordinate the system while recovering.
HomeRelay creates a clear plan, coordinates the right people, and keeps everyone in sync when it matters most.
Discharge instructions become structured tasks, medications, and follow-ups.
A personalized first-72-hours journey is created for the patient and family.
Caregivers, rides, and check-ins are assigned before something gets missed.
If instructions conflict, HomeRelay pauses and escalates instead of guessing.
HomeRelay coordinates decisions. It does not invent them. The plan stays blocked until the right human sees the cited evidence and answers.
Page 6
"Take lisinopril 10 mg once daily each morning."
Page 12
"STOP lisinopril. Review with cardiology at follow-up."
Medication blocker
Maya can send a prepared, cited question to Dr. Chen. Agents cannot choose which medication instruction is current.
HomeRelay turns vague family work into a shared plan: who is picking up medication, who is checking in, and what Robert sees next.

Omar
Assigned for Friday evening before Robert gets home.

Maya
Cardiology option and calendar artifact stay visible to the family.

Robert
One calm next step, not the whole coordination system.
HomeRelay is an MVP for the Band of Agents hackathon. It proves the coordinated Robert demo with real collaboration boundaries, cited data, and visible AI/ML API usage without pretending to be a hospital system.
Separate Band agent identities coordinate interpretation, medication, follow-up, caregiver, and review work.
Powers extraction, medication reconciliation, follow-up coordination, and structured safety review across the agents.
Rooms, tool results, citations, and human decisions remain visible as structured demo evidence.
This is a fictional hackathon scenario, not production clinical software or medical advice.
Robert, Maya, Omar, and Dr. Chen are seeded. The Band agent collaboration, AI/ML API workflows, citations, and handoff records are the real MVP work.