Also it will not prevent spam. So if you have a errornous device for example writing about 20 permanent hardware errors per second into your errpt, you will get a lot of mails.
Maybe a self scripted solution is better, just checking in let's say 15 minute intervals the entries of the errpt.
To expand on what zaxxon said: you can set up syslog facilities to collect the error messages you want to see in some file. write a script that parses this file and sends mails for certain circumstances to your designated email account.
For an HACMP admininstrator there is an easy way to add his own error notification methods.
# smitty cm_EN_menu
HACMP is the only AIX addon that comes with a SMIT panel to do this.
I use a free product called big brother to monitor all kinds of things on my Unix servers, it does have syslog integration so that you can parse words or phrases that you want to trigger alerts on. Check it out at Download - Big Brother System and Network Monitor and look at the 'btf' (Better than free) version.