Resource Management - Cloud Computing style

Tired of paying high fees on a per project basis, we were wondering how to reduce cost.

Though we hesitate to do IT project in-house - with the burden of having full time employees, location setup costs and higher on-site wages; getting the job shipped offshore was also problematic - due to lack of monitoring.

We tried some of our industry standard methods:

Fixed price - typically professionals would have issues about requirements at the end - and would speak of misreading/misunderstanding them at the time of delivery. At that point it would be back to the negotiating table.
Hourly rates - We found ourselves paying significantly higher amounts in this model - and started wondering if the same professionals slowed down? Or perhaps they were snoozing during billing hours?

One solution that came to mind was odesk:

odesk is a website that allows you to hire offshore workers, and monitor their work too. So you're sure you are only paying for real work time.

Meet Cloud Computing:

However, the best solution for us, came in the form of a former IT company we found on guru. We knew this firm to be professional in their work - committed to timelines and delivering quality code. But they were expensive.

To use their expertise, while cutting costs - we worked out a unique contract that would work for both of us:
We promised them a long term contract - and we would pay a fixed monthly charge per resource.
Based on the project, this person could be a designer, a programmer, or a tester.
As long as we do not exceed the 180 hours in the month, it won't matter which we use.

Next, each project (or 'work packet') we hand the company - would have a mutually agreed hourly effort.
This gave us the flexibility of assigning different types of projects to the company, while having a low blended rate. The company is also hedging its risk, given a long term contract.
The management and resource allocation issues are handed over to the company - so they can plan their resources given our work forecast.

Update:
So far, this plan has worked well for us. Of course, we're just testing the waters here - we have built flexibility into this plan, so we aren't too tough on hourly variations in effort for now.