My company develops embedded systems and they have a project where they are developing a new platform which will be used to do demos for potential clients. They are introducing a new technological concept and they want to have it as base for new products, so they can reuse most of the blocks of code generated and have just to configure some parts of the architecture and add customer applications.

There is a group of engineers working on it, a couple of them for each competency (SW/SYS/EE/Mech) and they want to include us (system testing group) to perform test over each release. My manager wants to include integration testing to our approach and I am leading the initiative.

I want to demostrate that we should only perform integration testing, since the software developed will not have an specified client, the system requirements will be generated by SYS/SW competencies. System testing is based on req. documents, so testing implementations done by SW against requirements written by SW/SYS will lead to issues not solved (I think SW will not put efforts to modify an implementation they decided to do). Also, most of the Requirements generated by them will be different that the generated by the potential client (in case they decided to give us a project) so that would only led to duplicate work.

Integration Testing (between software modules, not HW-SW) will lead us to find issues on the base architecture for future projects using that platform, and to find better ways to interface between applications the customer would want on the product and our sw and the provided by third parties (HAL, RTOS, Services, etc.).

I am right? What would be your criteria to decided whether Integration or System (or both)?