I'd recommend using Tryton instead, a real open source community project that forked from what is now called Odoo about 10 years ago. With Tryton you get upgrade scripts to keep an instance up to date, and all the functionality without any paywalls.
A law firm has started to use Tryton to manage their stuff some years ago (http://www.tryton.org/business-cases/advocate-consulting-legal-group.html). One of the lawyer there was responsible for the whole project, we helped him to understand mercurial, make patches and so on (I even wrote him a recommendation letter after he has proven his value so he could get in a university to get a CS degree).
So my advise would be that you should start to scratch your own itches and find some people to help you when you're stuck.
You could try https://github.com/odoo/odoo
The community version is limited. For inventory as long as you don't need barcodes, it could do exactly what you want.
I have never tried the following, but they are also available:
Well, if people believe the financial system works, then it works. Billions of dollars is not much, compared to the value of trust in the whole setup.
When your company is done throwing money out the windows (the ERP project, which will waste more company resources than you want to think about, because you will now deploy a version that is mostly written from scratch and completely untested, you will be beta testers and paying for it, and you know it), there is a free software ERP framework project that is fairly stable and mature. I am calling it ERP framework, because you will most likely need additional stuff programmed for each specific deployment. But since you are not paying for the license, you can afford a little extra. €20.000 license fees is more than a whole month of developer time.
The project is called Tryton.
We're working on Tryton (http://www.tryton.org/) a python framework to build business softwares (aka ERP) with batteries included (more than 100 modules).
It's used by GNU Health (http://health.gnu.org) which aims to provide digital health solutions with only free software. It's mainly used in third world countries.
That's an ambitious thing to embark on, and quite doable, though probably going to show up as more difficult than you intended.
Is this something where Tryton may be attempting to work?
I can help out a bit on mentoring for how to structure things and review if you want.
You could use Flask with flask-tryton extension. It will provide you all the needed tools from Tryton to manage sales (as it is a business solution aka ERP) but also all the flexibility from Flask.
Why not giving an opensource ERP like http://www.tryton.org a try ? You might want to start with a basic stock setup and then if the need is present you'll have the opportunity to install additional modules.
It depends by how we define "best". It has a lot of features, yes. A lot. But for a small firm, it might be overkill. As a python dev, if you end up helping your friend with installing Odoo, you might regret it because you'll find that working within its code base is not fun at all.
Also, with its new "proprietary" turn, I think its community is going to turn toxic and anybody not following in the "Enterprise" (proprietary) package is going to get burned. You don't want to assume maintainership of a fork of that shitty codebase, so you'll have to abide to whatever new scheme OpenERP S.A. comes up with.
I don't know erpnext, so I don't know if it's a better choice though...
EDIT: You might want to look out Tryton, a "community fork" of OpenERP made ages ago. I don't know if it's any good, but at least there isn't a proprietary Damocles sword hanging over it.
I would suggest Tryton. It is written in python so your IT department will probably have someone knowledgable in the language already. A risk with free ERP software is finding out it doesn't have the features or data you need.
Message me if you have questions about supply chains, eps, manufacturing, process engineering, ect...