Excel (and other spreadsheet packages like Calc) are good for non technical users it's useful for rapidly prototyping stuff but it is absolutely a pain for automation and any type of statistical heavy lifting. Once a spreadsheet reaches a certain level of complexity it becomes a pain to support and maintain much better to have a real database driving things.
I've heard a lot about R which is kind of an open-source version of SAS (at least the statistical parts) not sure what its automation and reporting support is like my limited understanding is it doesn't have the BI parts.
If Excel had a real scripting language as it's Macro language the problem wouldn't be nearly so bad. All that business code could just be copy/pasted, have some unit tests implemented then refactored so that the accountants code is now updated by the developer. As it is now the code has to completely rewritten in a new language.
As for the language choice, VB.Net or C# would be fine. I'm not saying it has to be Python. My complaint is that the code is non-transferrable. The apps cannot grow past Excel, and the email the most recent version to the next guy game gets you into versioning hell rather quickly.
He wants his pivot tables, KPIs and Solver and some macros. He wants point-and-click ETL. If it's a department/cross-department model, do it in SQL Server, or whatever else you use. You can use your IT kung-fu there. This is about what libre office lacks. Excel has power and flexibility and remains easy to use for non-programmers.
My issues with Excel come from the perspective of modifying these ad-hoc apps (many business run off these for a long time as they grow). Yes, there are other great features, but for me, for growth, Calc is taking a better approach.
Vendors include SAS, Oracle, IBM, Predixion.
Also Excel is used as a self-service BI tool. It has ETL, OLAP and dynamic dashobarding.