Erik Andersson is the CTO and co-founder of Carmen Systems. In this interview Erik shares his thoughts on Carmen's current R&D projects, including Data Management and the KPI project with MIT.

Optimization and Rave used to be your two key technologies, but now you talk about five different technologies. Aren’t you spreading too thin?
No. If you look at what we’re actually developing inside Carmen, we concentrate even more on the areas where we can make a difference. In the automation area we’re using Python – it’s an open source scripting environment. We don’t develop that. What we’ve done is that we’ve integrated with our optimization so clients can automate and shrink their planning process. Then we’ve integrated it with Rave and browser technology so clients can get interactive reports on the web, and eventually monitor and control their whole operation through a web browser.

So it’s more about connecting your core to> standard technologies like Python and the web?
I don’t mind if you’re putting it like that. Development wise, we might end up doing less. If Carmen reports are created in Rave-enabled web-design or publishing tools, we can decommission our proprietary report generator. Our employees do analysis and design, not printer-drivers and that sort of thing.

Does that mean that Carmen people are
becoming artists rather than developers?
Absolutely. I like to think I come to work and meet with artists. I guess it’s a bit of cultural thing at Carmen too. Good programs are beautiful, but when they’re combined in a bad way the design becomes inconsistent and ugly. Our clients are suffering from very ugly IT systems. We will replace them with more beauty.

But now you do data management as well. That can’t be all artwork. There must be a lot of profane work needed there. Isn’t there?
No. I think Dave (Data Access and Versioning Engine), is an excellent example of adding beauty. You can do new things with Dave, like exploring the entire planning process and learn when and why things went wrong. If it was mostly profane, we wouldn’t do it.

You’ve said that you don’t like financial systems, but now you develop financial indicators in your KPI project with MIT. What’s the difference?
My problem with financial systems is that you get the information about three months after the last time you could do something meaningful about it. We’d like to predict this, and next quarter’s financials and the impact of your actual decisions on that.

So you have to protect your system from inside traders?
Oh – you’re right, I didn’t think of that. If inside trading doesn’t become a problem, our KPI project has failed!