Many websites, including our own, can take a considerable amount of time to navigate to find what you are looking for. In this blog post I will tell you how to save time with one use case example. Because I have been working to structure the product data on our site, meaning machine-readable for Artificial Intelligence (AI) technologies, the end result is a much more pleasurable user experience and outperforms what are fast becoming the legacy internet search engines.
If you are like me when it comes to building models, particularly scratch building, I already have in mind a list of parts that I find easier to source as opposed to making the parts myself in the shop for building a model. However, model projects often go on hold until I can find what I need. Parts can be found at O Scale train shows, eBay or any other number of sources or simply trading with friends.
In order to cut that time down, it helps to have a part number which many suppliers like All Nation or Walthers would have in their catalogs with a description. The first thing you will need is a tool to facilitate the search on the Internet. Those tools would be a web browser and access to one of the AI tools known as ChatGPT or Copilot with the option to log in. For Copilot you will need to use the Microsoft Edge Browser, navigate over to Bing.com and then select Copilot to get started. Typically I log into Copilot before interacting with its user interface. This allows you to start communicating to this AI Large Language Model (LLM) trained beast.
Starting very generically, you could type in the following:
“O Scale 13 rung brass freight car ladder.”
If you have not logged in results might come from many different places and they may not make sense leading one to believe this does not work so well.
However, if you have logged in to be working with the AI model very interactively, then Copilot/ChatGPT or any particular AI engine model will get to know you and your interests. Copilot may sense your interest in the model railroading hobby and suppliers that interest you. Consequently, the results will likely be more accurate.
Copilot returned for me by specifically including a supplier name and thus a similar result:
“All Nation Line offers a pair of 13-rung brass ladders (PN#TC25) for $6.00. These are designed for O Scale freight cars and are currently in stock.”
Well, we got lucky here and it returned a part number and identified where to get the part. But suppose you only had an obscure part number such as PN#TC25 and you typed that into Google. Very likely you would not get a direct answer or if it does find something from All Nation, it is way down the list of results. If you were to type into Copilot just the part number like PN#10567 you might get really lucky. Copilot returned something similar to the following result as follows:
“PN#10567 is listed as a Whistle Chime (Brass) for steam engines, available for $4.00 on All Nation Line’s website.”
And sometimes if Copilot does not come up with a result for a part number you can implicitly tell it: “Part Number 3018 All Nation Line” and the result:
“PN#3018 from All Nation Line is a Brass Bulb Straight Brace, available in a set of 12 for $6.00. These brass detail parts are designed for structural reinforcement on prototype freight cars, adding depth and dimension to your model. They feature punched holes on each end for easy installation.”
So now for the real use case scenario using the list you have in mind to source all your parts. One can start with the following statement input into Copilot:
“I want to build a 1:48 scale model of an Outside Braced Box Car. Can you source the parts I need from All Nation Line?”
Copilot provide an answer with a pretty good suggested list which would have saved you a lot of time hunting for them on the website.
Outside Braced Boxcar Sub-assembly Detail Kits – PN#6550 ($30) and PN#6551 ($40).
Brass Boxcar Floor for 42′ Box Car – PN#539AN ($30).
Box Car Ends (4 Over 5, Riveted Sides, Metal) – PN#478AN ($6).
Roof Walkway Saddle Supports – PN#C933 ($5).
40′ or 50′ Metal Roof – PN#8510 ($7–$8).
Unfinished Embossed Wood Roofwalks (40′) – PN#3181A ($4).
Brass Fishbelly Underframe (1 Pair) – PN#3028Brass ($20).
Boxcar Accessory Kit – PN#2479 ($40).
Depending on your creativity in telling Copilot what you want and how you want it, you can actually filter with a different look so now one might tell Copilot to do the following:
“Can you put this list into an order form that I can send to All Nation Line?”
Now you can see the above product items in a table to scrutinize it further.
Now because I myself am lazy I asked an additional question to Copilot to give me prices and a total.
“Can you add to this order the prices and a total?”
Now what caught my eye was the total price and wondering why it is so high. Upon further scrutinization, I see the problem, so I need to filter out a couple of things I didn’t want and add a set of trucks.
“Can you remove part numbers 2479 and 6551 and add a pair of Bettendorf Trucks?”
Copilot then comes back with a more refined table as follows:
At this point you could pull it out of Copilot, pdf the order form and send it to us to make sure we have those parts in stock given that our inventory stock is constantly fluctuating and the LLM may need to learn what is currently In Stock.
If you were to do this same exercise without an All Nation Line Specification, you will get back a basic list showing some of the items found from All Nation but also some from eBay or another hobby outlet. Unfortunately, the returns from other 3rd party sites may be expired, sold or removed from inventory or were simply a 1 time offering in the secondary market expired. With the structuring of our product data, the precision about the results from an AI engine is superior to the average run of the mill Internet seach engines. The way we manage this information can be demonstrated in an example of a single product’s structure format that is only machine readable follows:
{
“@context”: “https://schema.org”,
“@type”: “Product”,
“sku”: “D5”,
“name”: “CB&Q, C&S Caboose Decal Set Burlington Route PN#D5”,
“availability”: “https://schema.org/InStock”,
“stock”: “8”,
“price”: “15”,
“priceCurrency”: “USD”
}
It should be noted in the above example use case, the availability or “In Stock” term is only an indicator that All Nation carries the part. It does not imply an actual Integer value that would be “Stock” meaning zero it is out of stock or say 10 units are Put To Stock ready to be shipped if ordered. Inventory levels constantly change, thus it is referential information only.
Simple yet highly effective results for the end customer. Thus, in the future I will be doing structured data representations on parts lists contained within say, a gondola kit or a trolley car kit, to make the AI models even smarter. Consequently, you may have just realized it only took 5 minutes of your time to source the list of parts, and perhaps the ones that are missing might not be out on the website yet if it is an older part number.
While all this machine readable data is nice and it is hiding under the covers of the web, it can be visualized in a graph thus giving one some indication of the relationships within the data structure.
Now let’s get started and try this out!
