Back when I used to live in merry England, I worked in a consultancy. Yup, I was a vendor. I used to love doing demos – it was an opportunity to showcase our product in a controlled and pristine environment where it worked perfectly. Pathways and navigation where predetermined, search results perfect, flawless functionality. A bit like taking your kids to someone’s house and they behave faultlessly. Client questions were easily answered or deflected. Naturally, it helped that we had a great product. ‘Demo Death’ was a rarity.
Now that I’ve moved over to the light, I love vendor demos. Its an opportunity to pick their product apart, puncture the hermetically sealed environment, divert from the pre-ordained pathway and functionality and see what the product is really like.
Having been on the other side, I know it is unlikely that the product will ever work quite like the demo shows. But that’s ok, sometimes you don’t need all the bells and whistles, you just need core functionality.
I have two favourite questions: ‘Is that functionality ‘out of the box?’ Translation: Am I going to have to pay for some extra services? and What percentage of your revenues are sales and what percentage are professional services? (Translation: If there’s a high percentage of professional services, much of what I just saw is not out of the box and you guys are going to get rich with the services I’ll need to sign up for. I think I’m pretty good at seeing through the smoke and mirrors of a demo.
I went to a product demo yesterday that was impressive in the extreme – one of those demos where you hand over some of your content a couple of weeks before and they apply it to their product and generate a set of ‘results’. Except there was no way to fake the results I saw yesterday, it was just too complex.
The company was Idée Inc. If you’ve ever used the Visual Search in Photoshop Elements, that’s their product. But they also provide a rights management service. You give them some images and they tell you where they’re being used in print and online. Yup, that’s right, ‘print’.
We sent them about 500 images from our permanent collection and they demoed the results. Spectacular. To give an example, they found matches in Home & Garden magazine which had pictures of interiors where the poster on the wall was one of ours. It found details of images within montages, it found personal websites where people had taken their own photos of our works, it found skewed versions of images, black & white versions, doctored versions and background webpage images, it found book covers – it apparently has no trouble with overlayed type. There were hundreds upon hundreds of matches – no way to fake that.
The thesis for my postgraduate degree was on the development of image recognition software, so I like to think I understand the complexity of what they’re doing but I’m hard-pressed to grasp the scale and accuracy of it. In terms of target source, they’re scanning print media on a large scale and they’re buying 3rd-party crawl lists for the online piece. They must be reducing our images to a vectorised or fractal format, doing a proximity match on the target source, then comparing the details based on a proximity threshold. Alternatively, it could just be magic.
I think we’re the first museum to talk to them so we don’t fit into their regular business model and my last question to them was if they were a publicly traded company.
I’m going to have to classify this product using the British vernacular reserved for something extraordinary: The Dog’s Bollocks…