Notes from the road  

A few years ago, the Gnosis team started a tradition of driving an RV cross country from our headquarters in Charleston, SC to the JOC’s TPM conference in Long Beach, California. Along the way we stopped to get face time with our customers, onsite at their plants, factories, warehouses, and offices. One year we even recorded our journey in a documentary short called Between the Ships.  

While the RV stayed home, September brought our team back on the road.  

Over the course of ten days, Gnosis teams moved through three industry events: the North American Supply Chain Executive Summit in Las Vegas, Automotive Logistics & Supply Chain Global in Plymouth, Michigan, and JOC Inland26 in Chicago.

Here’s what this season of travel confirmed for our teams: freight becoming more digital does not make it any less physical.  

The containers, railcars, trucks, parts, people, and handoffs are not in line to become obsolete. Software can make that network easier to see, understand, and act on—autonomously even in some cases. But to build good software for freight, staying close to the people operating freight is non-negotiable.  

So we go where they go.  

Las Vegas, NV  

We started at NASCES in Las Vegas.

There was, to no one’s surprise, plenty of AI in the programming. But what was interesting was a noticeable impatience with talking about AI in the abstract. The better conversations were about what the technology should accomplish practically.  

What should AI actually do? Where does it fit alongside systems like SAP? What is worth automating? What is not? And what outcome should be different when the project is finished?

One speaker made the point that the hardest part of a technology transformation is not necessarily assembling the people, systems, or processes. It is having a clear enough vision of what all of them are supposed to achieve.

That showed up in our own conversations, too. People responded less to AI as a category than to seeing it attached to a specific operational outcome.

This felt very much aligned with the order of operations we believe in here at Gnosis: Start with the work. Then decide where the technology belongs.

Chris Nielsen of Toyota made a related point from another direction: a supply chain is not improved simply by squeezing every vendor as a cost line. Relationships helped Toyota when it needed them most. A transformation still must work through the network of people and companies carrying it out.

Even the best technology eventually must survive contact with the operation.

Plymouth, MI

Second stop was the Automotive Logistics & Supply Chain Global conference in Plymouth, MI.  

If Las Vegas kept asking what technology should do, Plymouth supplied a lot of the underlying conditions it must work with.

Data came up constantly. As one speaker put it, “everything starts with data.” But knowing where the data lives and how to access it was a little less ubiquitous.  

Many companies at the show were in the middle of change: mergers, WMS rollouts, combined networks, provider consolidation, carrier benchmarking, and efforts to use historical data to catch problems earlier.

All of that depends on having a coherent picture of what is truly happening.

That is easier in some parts of the supply chain than others. Rail was identified as the problem child—“black hole” were the words several people used. Some shippers we spoke with are still going directly to individual carrier websites to figure out where freight is. Meanwhile, those same organizations may be rolling out sophisticated systems elsewhere in the business.

That is less a contradiction than a feature of a network built over decades across carriers, modes, facilities, and companies that were never designed around one shared version of the truth.

Companies are trimming partner lists, but they are asking more of the providers that remain. That means better KPIs, clearer accountability, stronger historical performance data, and a better understanding of where each partner performs well.

Visibility and partner performance are really two sides of the same problem. You cannot meaningfully evaluate a provider if you cannot reconstruct what happened across the movement in the first place.

Chicago, IL

We finished September at JOC Inland26 in Chicago.

By then, a question from the previous two shows had become more specific: if we know what work we want technology to do, and we know the underlying data can be messy, how much responsibility should we hand over to the machine?

Our CTO, Jake Hoffman, joined a panel tasked with answering these very questions in the session titled “Human Out of the Loop?”

Jake drew a practical line between work people probably should not spend their time doing anymore and work where human judgment still matters. Taking information from a PDF, spreadsheet, or email and entering it somewhere else is a good candidate for automation. Setting goals, defining constraints, and deciding what a system should optimize for are different responsibilities.

That makes autonomy partly a question of context and trust.

Jake described the path as incremental: verify the output, understand where the system fails, make sure it is doing the work correctly, then expand what you are willing to entrust to it.

That same week, our Product Owner, Jack Hixson, recorded an executive interview about forward-deployed engineering at Gnosis.

It is a useful companion to the panel because it gets at where that context comes from in the first place. Before we can automate an operation well, we need to understand the human actions and nuance that govern the process.

Some of that knowledge lives in systems. Some of it lives in the exception somebody knows to watch for, the macro-enabled spreadsheet everyone depends on, the process that only makes sense because of something that happened two years ago, or the person who knows who to call when a particular lane goes sideways.

Forward-deployed engineering is partly about excavating that tribal knowledge and turning it into something the software can use.

The rest of JOC supplied plenty of reminders that the work itself remains stubbornly physical: cargo theft, fuel, intermodal capacity, carrier performance, volatility, procurement.

A couple of lines made it into our notes:

1. Lowest cost isn’t always the lowest cost.

2. Redundancy is not resilience.

Both were really about the same thing: a supply chain looks different when you optimize for what happens collectively in reality, rather than what looks best in isolation.

What comes back with us

Of course conferences put customers, prospects, and partners in the same place. But they also compress a lot of an industry into a few rooms.

Put enough of an industry into a few rooms and you get to hear how people describe the work when they are not answering a formal discovery question. You notice the same phrase coming from companies that have never spoken to each other. You hear the polished version of a transformation from the stage, then the operational version in line at the open bar.

For us, that is part of product work. It is another way of collecting the context that makes the product useful.

The RV trip to TPM was one version of it. Forward-deployed engineers embedding with customers are another. Three conferences in ten days are another still.

Freight happens out in the world.

It makes sense that some of the knowledge required to build for it has to be acquired there, too.

Which is how our run ended in Chicago with our friends at IMC, seafood towers on the table, football on the big screen, and a Bears win narrow enough to keep dinner interesting.

Not everything worth bringing home fits neatly into a CRM field.

Some of it ends up in the product.

Some of it ends up in a notebook.

And apparently some of it ends up in a highball glass.

Appendix:

The O’Keefe

One of our AEs insisted everybody order this. We are documenting the recipe in the interest of historical accuracy.

- 1.5 oz Fernet or Montenegro

- 0.5–1 oz espresso

- Top with tonic water

- Orange twist

- Highball glass

Build over ice. Interpret “0.5–1 oz espresso” according to how the day went.

Continued Reading

The Age of Wood — A. R. Ennos

Guns, Germs, and Steel — Jared Diamond

‍