As someone who has worked on larger enterprise stuff this would be incredibly (like a salary a year incredibly) useful tool.
It would do in software what a really good reliability engineer with years of knowledge can do "Foo has broken" "Nah foo is fine, someone changed Bar again without reading the wiki" stuff.
It's funny, the reactions I get. A lot of startup programmers who have never worked on large systems don't understand why it's so valuable. But those who have done big enterprise use phrases like "holy grail".
But this is a HARD problem. Man, abstracting change in an actionable, readable, efficient way has been difficult. Not to mention the security issues and other headaches.
I'll bet, it's a truly monumental task to take on (so kudos for giving it a try), stuff I work on these days doesn't need that level of monitoring (We have 4 servers and a bunch of desktops laptops, it's pretty much vanilla small business stuff) but I know a few people who would love that kind of stuff.
opsmatic.com ?
A different focus, but same basic idea. ScriptRock, too. I met the founders of both of those companies at a conference last fall, and it was really interesting. The products diverge a lot, but we all see the same fundamental problem.
The product is a way to answer the simple yet awful question, "What changed?" Configuration change is the primary cause of failure in complex software systems. But the part that breaks isn't necessarily the part that changes. For example, a database schema change can break apps that rely on the database. But in large organizations, where the database is administrated by an entirely different team from app development and developers may not even be allowed to look at the production environment, it's a huge pain point.
Detecting configuration change and finding the relationships and impact is a really difficult and challenging problem. The biggest takeaway from the alpha for me was that "noise shaping" is a hard problem. For example, Wordpress alters the schema of its MySQL database several times a day. It's a tiny thing (an autoincrement value on a single column), but it makes sorting important changes like upgrades that only happen weekly or monthly or yearly from unimportant changes that happen daily.
The trick now is to get a product out of it good enough to make money, so I can afford to quit dayjobbing and live off my own creation, rather than sitting at someone else's desk and counting the number of times a day I wish I had my own product here to help solve the problems I deal with over and over, every damned day.