Productivity · News · 🇧🇷 Brazil
Hospital predicts discharges with AI and gains the equivalent of 12 beds without construction
Hcor, in São Paulo, built Command Flow, which estimates when each patient will be discharged and coordinates room turnover.
- Profession
- Hospital Manager, IT Professional
- Tools
- Command Flow
- Country
- 🇧🇷 Brazil
- Report date
- October 22, 2026
- Verified on
- October 27, 2026
What they did
The technology team found that most delays in freeing up rooms came from communication gaps and pending paperwork. In four months, in-house teams built a command center that estimates discharge six hours after admission, recalculates at 24 and 48 hours, and coordinates steps such as room cleaning and the arrival of patients waiting for a bed.
How it went
According to the hospital, the third prediction gets 92% of discharges right, and the faster turnover is equivalent to 12 to 14 beds. One month after launch in January 2026, Hcor hit two targets for the first time: 60% of beds freed within 60 minutes and 60% of discharges by 10 a.m. The investment was about R$ 1 million; the R$ 36 million mentioned is potential revenue, not realized revenue.
Good to know
The prediction relies on medical record data, which is sensitive under Brazil's LGPD. Treat the estimate as support for operational planning: the discharge decision still belongs to the physician.
The prompt
I manage [hospital or clinic] with [number] beds. Using this month of data [table with time of medical discharge, patient departure, cleaning and next admission], show where beds sit idle, which steps cause the most delay, and propose 3 targets, such as [bed freed within X minutes].
Replace what is in [brackets]. Prompt written by us from the account, not a literal copy of the source. Do not paste other people’s personal data (privacy laws such as LGPD and GDPR apply).
Summary in our own words. Results as stated in the original account; we have not tested or audited them. Read the source ↗ Found a mistake?
Similar cases
01Daily newsletters become a podcast at 5 a.m.
The agent reads the newsletters under a label, skips the weak ones, summarizes the rest and generates audio with a local voice.
02Two HR staff build an AI performance review platform without knowing how to code
At Riachuelo, two HR employees built a tool in 12 days that automates steps of the performance review cycle.
03Claude doubled the speed of a home Wi-Fi network
A user asked for help with slow internet; the agent ran tests, read the Mac's signal data and walked them through the router settings.
04Product manager turns scattered notes into a priority list
Ron Forbes dumps meeting notes into a folder and Claude Code returns summaries and tasks sorted by priority.
05A table at a hard-to-book restaurant without refreshing the website
The user asked the concierge to watch for an opening at Don Angie; the agent booked it as soon as one appeared.
06Check engine light on: agent bought the scanner and found the fault
A driver told the agent about the dashboard light; it ordered an OBD2 reader, read the code and cleared the alert.
07One week to find an apartment in San Francisco
A new hire used Claude and ChatGPT to list property managers and filter listings by price and area.
08Private chef and ski trip arranged during a dog walk
Paul Metcalfe gives spoken instructions and ChatGPT requests quotes, reads replies and summarizes the options.