Common Object Management/Data Access Methods

Folks, I'd appreciate a thoughtful but rapid response:

Has anyone had experience with using "objects" to create a layer around
their databases, a Common Object Module(COM) as it were, that supposedly
makes access to databases easier for end-users and applications
developers alike.   We have been approached by a vendor that says he has
built and can build around each of our divergent databases within
different database machines, a "data access COM" made up of objects
built on "object technology" similar to CORBA or Microsoft's derivative,
that will contain the necessary data access methods and logic able to
mask having to know the underlying data structures of any database and
any complex operations of any application front-ends now existing for
applications against those databases.   Once built, this COM, he claims,
 would make it be much easier and faster for applications and queries to
be built.   I'm skeptical and hear "silver bullet" talk but I'm willing
to be convinced.    

If anyone has had such an experience, how did the objects work, how
were they built, how complex a task was that, what languages and data
access methods are usually involved, what kinds of resources and skills
were involved, roughly, how long does each "object" take to create, what
are the support, performance, security, and management ramifications,
what is the "good news, bad news" of which someone should be mindful?  
I did hear ODBC mentioned by him, which gave me shivers.

Any other thoughts?

Thanks in advance,

Dave 
Arizona