Practical use of AI in FreightTech - world's first process?!
In previous editions we talked a lot about why the speed of AI could be MUCH faster than previous technology waves.
We also wanted to temper expectations to say that there is a lot of hype around AI and it could be "a hammer looking for nails" in some scenarios.
So for this edition, we are going to dive a little deeper and go into a practical example here at Ship Angel. Where could AI make an outsized difference right away, if at all?
We started writing code in February for the first version with a go-live goal in August. We had lots of input where AI could be included in the product; one of the favorites was a "co-pilot style" capability to suggest the best vendor for each shipment based on many factors. However, this would not add immediate value, mainly because it needed to learn - which takes time...
领英推荐
Then we found one! Previous generation Rate Management Software relies heavily on scripts and BPO centers in India and the Philippines to manually transform rates from individual vendor formats to their standard. This is costly and error-prone. So we developed our own Seven Pillar ML/AI Architecture (the picture attached to this article is a high-level view of this) to read and transform all the data on rate sheets and push it into our system/API.
While the priority was to go live, our small incredibly talented engineering team worked on this in tandem, and recently got it working with a number of the large shipping vendors' rate sheets. This is a good example of AI making a real impact fast, and possibly something that has not been done before (large rate sheet ingested and transformed without human touch/input)?!
If you're a customer or potential customer, we actually include this now in the product demo to show you how it works!
To conclude, this is a real-world example of AI making a significant difference in a product fast and early, so it can be done... BUT it doesn't plug directly into every problem today, based on the maturity of the ecosystem. I guess my main advice is to concentrate on areas where it can add value fast.