What was best before sliced bread?

| |  ← bread sliced by machine.
| \  ← bread sliced by me. 

My toaster does great with pre-sliced bread, but when I insert my raggedy irregular triangular prismatoid of a hand-slice, it burns on one side while the other remains as pale as European renditions of Jesus (although he’s never shown up on my toast). Then the whole slice inevitably gets stuck mid-pop and I have to dig it out like I’m excavating ancient Roman artifacts¹.


It was actually the home pop-up toaster of 1926 that really drove the demand for pre-sliced bread. Toasters were highly useful, but hand-cut bread slices were so inconsistent that it was hard to get consistent results. The first commercial slicing systems arrived two years later in 1928,  allowing people to easily and repeatedly make decent toast. 

A new technology shaping how an existing product is made or used is a common historical pattern. Could generative AI be the pop-up toaster to the bread of our customer support? We’re already seeing help documentation being written to make it easier for AI tools to imbibe, in the same way web pages have been sculpted to suit search engine retrieval. 

The support teams we built over the last couple of decades were shaped mostly to the strengths and limitations of people. The teams we’ll build in the coming decades will integrate the capabilities and limits of AI tools, and human roles will be reshaped alongside them.

Many people were already good at slicing bread — toasters were useful for them immediately. Pre-sliced bread made toasters more accessible and useful for everyone, including the slicing-challenged. It’s what you do with your toast that makes all the difference. 

You’re already good at support, but new tools will make some of your work easier, and also make some of it more accessible to new people coming in. Perhaps you’ll always prefer to hand-slice, but it’s likely that pre-sliced support tools will become the new default option. 

The support content we create will now be made not just with human customers and support staff in mind, but with AI as a primary consumer. 

The types and timing of our support offerings will be reshaped according to the new capabilities AI makes available. 

Even as we adapt, the things that differentiate great support from mediocre support will still be what you do with those tools, and how you build systems around them. 

Human judgement, human quality control, human intentionality will matter more than ever when automation creates a shared basic level of service. 

 What are you going to put on your pre-sliced support toast?

¹ The Help Scout retreat was in Rome last week. It may have seeped into my writing a little. What a city.