Denise Roebell, director of the Garfield Community Physicians Group (GCPG), had just finished the notes from her Friday morning clinic and, at 8 p.m., turned to what felt like a never-ending stream of emails. She understood all too well what “pajama time” meant: the administrative work that so many medical professionals had to do late into the night. It was also one of the main causes of burnout, a major expense for a health care organization and one Roebell hoped to reduce.

She considered an email from the GCPG regional medical directors, proposing hiring medical scribes to help the clinicians with documentation, but this would be expensive and likely short-term. Roebell had also been reading about some new AI solutions for documentation, but she had concerns. Would providers, nursing staff, and patients feel comfortable having AI software listening in and writing notes? She had also heard about AI hallucinations and was concerned about nonsense ending up in medical records. Was the efficacy of these solutions worth the cost? She knew, however, as she finished her own pajama time, that the system was at a breaking point, and she needed to do something.

This fictionalized case is an abridged version of UVA-OM-1809. It offers a brief overview of electronic health record systems and introduces two AI products for medical documentation.