I have been experimenting with application modernization via unattended GAI agents. Using Codex, unattended, I ran an agent uninterrupted for 9 days with the task to modernize OpenEMR to a completely new Tech stack:
AngularJS → React
PHP (CSR) → .Net Core (.Net 10) (SSR)
MariaDb → Postgres
This also removed all stored procedures. At the end of this I then ran Claude Code against the new Public Web API and had it take a stab at building a new UI, which it did. The coding agents estimate that there is 95% functional parity.
I am interested in sharing the results of this experiment as I think it shows significant changes in how software will be developed and maintained in the future.
This is an experiment and I have no plans to compete with the existing OpenEMR solution or do anything commercial with this.
However, I also want to respect the licensing for OpenEMR. My experiment used the source code of OpenEMR as the solution specification, the reference material. It did not translate the code, instead it iterated across hundreds of small features and implemented each feature cleanly. So the resulting solution is an entirely new database and codebase.
Can anyone give me guidance here on how to share my experiment publicly without violating the OpenEMR license or irritating the community (this is not my objective). I am happy to use whatever open source license is necessary as I have no intention of promoting this, or even maintaining this going forward.
I have just made a quick video of this. If I don’t get any pushback I plan to make my github repo for this public in another week. OpenERM automated modernization
Thanks for sharing your experiment and checking in with us before publishing @Neil_Kimber .
Could you rename the app, the repo, and the video so they don’t carry the OpenEMR name please? Describing the relationship with something like “an experimental EHR rebuilt from OpenEMR’s feature set using unattended coding agents” is fine.
Trademark and copyright are different rights, so a GPL release resolves the code question but not the name.
Many thanks for the feedback. I’ll name it something entirely different to prevent any confusion, but continue to attribute back to OpenEMR. I’ll let you know when I make it public. If you see anything that concerns you let me know and I’ll change things to your satisfaction. I need to clone the repo and remove my migration process IP before I publish.
will you put the results on github as opensource for people to come in and test and improve.
that was crazy! did you work on it further?
there is a lot of modules that could be built in , ePrescription ; Clearinghouse, Lab results , telehealth ; medication history retrieval.
also ONC certification that openemr got since version 8
the problem is now codex is your programer and you dont have a community to work on it
security of needs to be audited perhaps
I would try to make it copy functions of Epic EMR as most hospitals are moving to and most providers are used to.
I am a US physician/nerd, and as much as I’d like to stick it to the big EMRs, OpenEMR in its current state is just too clunky and ugly to be a practical tool. It’s a shame because I know its capabilities and the work that has gone into it are incredible.
This, on the other hand, actually looks a lot more like a modern EMR, and the patient-facing side is palatable for a real practice. I could see myself using it, on the surface. Of course, the problem is we don’t know what AI-generated security vulnerabilities and bugs lurk beneath the hood, and that’s a BIG issue to solve. I’m not really sure the best way to go about that.
At the very least, this modern UI is an enormous upgrade and probably more innocuous from a security perspective. I haven’t looked at the code - could it be adopted by OpenEMR?
For everyone in the community information, I have started a modernization initiative where I’m currently consulting with maintainers and foundation board.
The initiative is focused on exactly the same issues I keep seeing on the forum. For those that don’t know me all you have to do is research me in core Git repo or contributions from a general search on this form. That should tell you I start and complete projects and this initiative is very important to me.
I will be doing a Community introduction and outreach for engineering help very soon so stay tuned.
By the way, I will take a closer look at this GAI project that everyone seems to like and give my honest opinion if it could be integrated with OpenEMR core.