AES will be selling a product that requires relatively rare, specialist manufacturing; they plan to closely match production to orders, ideally making products only when there are orders for them. In this way they are similar to fully integrated, made-to-order firms like The Cambridge Satchel Company, whose factory belongs to them and is right next door to their headquarters (though AES do not intend to own a factory since that is not their competitive advantage). My first instinct was to advise that they use an off-the-shelf e-commerce solution managed by someone else, such as Shopify or a managed Magento installation, but as we discussed the puzzles inherent in their ordering model it became clear that a standard tool is less likely to be suitable - a similar conclusion to the one we drew at my former employer Secretsales, where the rapidly changing stock on sale meant we needed our own software to manage copy, images, and orders.
The founders asked how they could attract developers to build what seemed to be a relatively less-challenging or interesting technology product. I initially suggested that the difficulties are all in the ordering and that they consider focussing on the technical challenges here, perhaps integrating other components off the shelf initially. As usual, the first problem is getting the initial developer in the door to interview all the others; here a consulting CTO like Andy Skipper can often be helpful as an initial catalyst.
However, at some point it struck both of us that actually AES doesn't need any substantial technology to start, as they just need to prove their product attracts customers at low volumes. (See Do Things That Don't Scale by Paul Graham for examples like AirBnB knocking on doors to find initial customers.) Something as simple as taking a customer's Emailand product desired, or using a Google Form to feed a spreadsheet, would be sufficient to register interest and (manually) trigger production. They plan to try this for the initial trial period and then build specialist ordering software only when scale actually demands it. Let's see how that works out!