How TNX leverages data

where AI sits in the UX

Our AI Pipline

In general, the software uses AI in the procurement process to automate pricing & tendering decisions. If you imagined AI as a very smart black box, the platform would look like this.

AI as a Microservice

But obviously you want to know what is in the AI black box. The diagram below shows how the platform is architected. We call this our AI pipeline, because it shows how data comes into the platform, is progressively treated with different AI techniques, and ends with tenders to carriers. Also note the feedback loops embodied in the process. our AI architecture

Each of the boxes in blue is a specific data science tool or process which adds value. The first step is to profile the approved carriers, understanding what they will be interested in and how this intersects with work they might be offered. Profiling combines 3rd party data, historical data, and the carrier's own reactions to offers made to them on an ongoing basis. Our carrier profiling is described in detail here.

Second, the software looks for logically superior ways to use a carrier's trucks. It ends with what we call “bundles”, which are specific ways to schedule, consolidate, sequence, and therefore execute a combination of loads. Bundles can be FTL multistop combinations, or LTL consolidations. Obviously most loads won't be bundled, so 80% or more of loads are tendered out in the same way they appeared originally in the TMS: as a single pickup and delivery. But when bundling works, the savings are substantial. Our bundling is described in detail here.

Third, we predict prices for every single load or bundle. This prediction is dynamic and based on a combination of historical load data, carrier profiling, and 3rd party market indexing. Our price predictions are described in detail here.

Finally, we select a strategy of how to procure capacity for the single load or bundle. A strategy means a specific set of tactical moves around the carrier inclusion, price negotiations, and timing of offers. whereas bundling is about objectively better use of vehicles, tender strategy is about subjectively attractive offers to carriers that achieve the best margin possible for the broker. Our tender strategies are described in detail here.

Our AI Methodologies

For the technical visitor who wants to know more, the software applies two AI methodologies to the needs of trucking. The first approach is what is known as statistical AI, or more popularly as machine learning. It is premised on the idea that a large volume of historical, current, and future data has patterns of importance. The platform discovers those patterns, and uses them to predict key outcomes. It is called machine learning because greater experience improves the predictive power of the platform.

The second approach is known as symbolic AI. Symbolic AI does not necessarily require learning from experience (although that can help). Instead of focusing on learning from experience, symbolic AI tries to accurately describe the state of the world, the actions available, and goals we want to achieve. With that, the AI acts as a rational agent trying to achieve goals with allowed decisions and with an expectation of how the world will react to them.

The symbolic AI side is mostly focused on bundling, while various forms of machine learning (reinforcement learning and supervised learning) are using in carrier profiling, price predictions, and tender strategy.

All of these AI methods are applied “in the middle”, between the load coming to the software and the tender given to the carrier. If the software does its work right, the reps and carriers never see the complexity of the AI, just the quality of its results.