This is probably the best Product Management approach I've heard of for early startups: http://www.quora.com/Stripe-company/Does-Stripe-have-product...
For most startups the PM role is fulfilled by the founder's vision. However, as the company begins to grow and the initial vision is achieved, it becomes harder to go without someone focusing on the product aspect.
The reason someone like Stripe can do just fine without PMs is for two main reasons: 1)Incredible market pull - they just need to be methodical about following the requests and prioritizing them in accordance with revenue/implementation sequencing/risks. 2)Stripe's engineers are the target market - Stripe is built by developers for developers to make payments easy. Much like Jobs embodied his own target market, whenever you are building something for yourself, market research becomes a bit less important... more implicit.
If the above conditions are true for your startup, then absolutely, you can follow the example. However, if you are missing either, you should really consider dedicating someone to this research and visioning.
It also seems like Ries was trying to direct this post at all kinds of companies, not just early startups. Patrick himself said that Stripe's approach to staffing is an example of "doing unscalable things while you can". Not everyone is a small startup and the approach of letting everyone manage themselves doesn't work at 100+, 500+, 1000+, 10,000+ person companies.
It all depends on on this:
>The engineers who build our products enjoy, and are good at, figuring out what the product should do—not just how the implementation should work.
you mention "early", this is key. PMless agile is super simple until you have live, paying customers (as in corporations, not people). breaking existing stuff is simply not possible anymore, SLAs are in play, validation efforts, etc.
throwing shit at the wall and see what sticks works in hardcore startup mode before you have a customer base. then it becomes suicidal. even in consumer space, see the various failed relaunches and redesigns.