Hi HN, I'm Sergey. I've spent years around ERPs where the useful data sits in places
nobody dares to touch: an AS/400 file in EBCDIC, a WinDev export where every column
is text, a CRM that only talks SOAP.
lumnik reads that stuff read-only and puts it in Postgres. You describe sources and
rules in YAML. It can merge two systems into one customer and keep both values when
they disagree, and it flags things like an order that hasn't moved in 30 days.
Every row is counted. If a line can't be read it's kept aside with the reason, it
doesn't just vanish.
Try it: clone the repo and run ./deploy/selfhost/up.sh. The output in the README is
from the demo.
The open part is Apache 2.0. The natural language questions are in a paid edition, not in this repo.
Hi HN, I'm Sergey. I've spent years around ERPs where the useful data sits in places nobody dares to touch: an AS/400 file in EBCDIC, a WinDev export where every column is text, a CRM that only talks SOAP.
lumnik reads that stuff read-only and puts it in Postgres. You describe sources and rules in YAML. It can merge two systems into one customer and keep both values when they disagree, and it flags things like an order that hasn't moved in 30 days.
Every row is counted. If a line can't be read it's kept aside with the reason, it doesn't just vanish.
Try it: clone the repo and run ./deploy/selfhost/up.sh. The output in the README is from the demo.
The open part is Apache 2.0. The natural language questions are in a paid edition, not in this repo.
Do you want to try a legacy source to connect in?