What we do:
We're building a community of vehicle data (http://shadenut.com). Mechanics and DIYers will be able to look up any piece of information they need to work on their car directly from their phone while still under the hood (ex. torque specs, TSBs, fluid types/capacities, etc). As a developer I've seen the positive effect that StackOverflow has had on our industry as a knowledge base and am trying to do the same for the automotive industry. The data will be crowdsourced and 100% free to use.
Our Problem:
Our biggest problem is that the product is not really usable unless it contains EVERY piece of technical data about a model, after which it becomes tremendously useful. The most common feedback we get from technicians is that they'd love to use it and contribute once it's a complete database (as long as it's accurate) but wouldn't switch from the paid competitors until then. The data is all available but there's simply too much of it for a small team to manually import. Our current strategy is to start with a select few models and incentive technicians to make their own entries.
However, I'd love to hear how Kevin and Sam would solve this or from others in the HN community who have faced similar problems.