Ecomm leaders on how to weave AI and automation into your processes
![](https://cdn.prod.website-files.com/663559c9207f91d5a1ba91e4/6660e4c5ca66498e662c0189_AI%20in%20supply%20chain.png)
Last week, Parabola’s CEO Alex Yaseen sat down for a conversation with Aaron Alpeter, the founder of Izba, a consulting service that helps ecommerce brands manage their supply chains.
They covered a ton of ground, from musing about a hilarious real-world excel macro that a company referred to as its “Craig’s list” to the journey of going from a solo-preneur to an entrepreneur. But ultimately, their conversation really pivoted around best practices for using automation to improve processes — and also an analysis of how AI is going to slot into improving ecommerce teams and processes.
Here are three takeaways that I think will make you sound smart at a dinner party if your guests also all work in operations, logistics, or supply chain…which I’m sure they do.
Your great-grandma is probably a knowledge point person — and what that has to do with building better processes
For Yaseen, Parabola’s users occupy a really unique space within their organization as what he calls “knowledge point people.”
“This is somebody who knows all the ins and the outs of the problems in their business,” Yaseen explains. “They know where all the gaps are, where all the skeletons are buried, and they have tons of ideas for what they or their teams could be doing to move their companies forward — but they don’t usually have the ability to write code or the technical resources necessary to implement their ideas. They’re just stuck doing things in spreadsheets.”
And these knowledge point people are absolute sponges for the processes that run within their companies.
Just like your great-grandma.
Is there something that your mom does that you’ve always thought was a little strange? For Alpeter, he calls to mind a story where a little girl watches her mom cut off the end of a pot roast before cooking it. When she asks her why, her mom replies that’s what her mother does. The little girl goes on to ask her grandma the same thing only to hear, again, “well, that’s what my mother did.” With her great-grandmother still alive, the little girl had a final matriarch to ask: Why do you cut off the end of a pot roast?
“The pan is too small,” she replied.
So how is this relevant to building smart, scalable ecommerce processes built for the long haul?
Despite not knowing why generations of women in this proverbial family were cutting the end off their pot roast because, they assumed there was a good reason: It’s what their mom did, after all! But without actually understanding the logic, they had adopted a process that was wasteful and unnecessary. The knowledge point person (great-grandma) had never vocalized or documented the process for making pot roast.
“In the context of ecommerce companies, there’s a lot of turnover — especially in these operational roles — and a lot of the time, the person who made a certain decision not only no longer works at the company, but maybe hasn’t worked there in years. So much of this knowledge gets lost, so it can be really hard to make changes to your processes,” Yaseen connects.
“This is why you really have to understand the ‘why’ behind things,” he continues. “To make a really good change to a process, you should understand the ‘why’ behind what you want to do now and the ‘why’ behind the historical decision that built the process you’re working to improve.”
How do you do this? Write things down, document them, and do whatever it takes to get them out in the open.
To outsource…or not to outsource
When it becomes clear you’ve got processes ripe for automation — and you’re really clear on the “why” behind it all — you’ll encounter the next question: Should we tackle this internally? Or should we outsource it?
Alpeter and Yaseen landed on a pretty solid framework for how to solve this (sometimes hairy) question. Here are some things you can start asking yourself:
- Is the goal to connect two systems and never think about it again? Yaseen recommends a lightweight automation tool like Zapier.
- Does the process you’re trying to automate feel really unique to you as a company? Is there a lot of complexity baked into it? If the answer is yes, you’ve got a couple more questions to think through.
- Do you have the internal knowledge on your team and pretty much know what you want to do? You just need to execute?
- Do you have no idea what needs to happen and really could use outside expertise?
If you don’t have anyone on your team that truly knows how to do the thing, you’ll probably need to get an outside expert to come in and help you learn how to stand up the process. This will ensure you benefit from the best practices and learnings that someone who has done work like this before can offer.
“If you have the existing knowledge internally, you already have the great-grandma who knows how the roast was cut, you may just need to get that knowledge that’s locked away in their head out into the world,” Yaseen says.
For Alpeter, this is mostly true. “I would also look at not only can someone do it, but when was the last time that they actually did,” he said. “You may have someone who’s perfectly capable of running with this but they may only do it every three to five years — someone who is able to do something is different from whether or not they are sharp and ready to.”
Plus some additional thoughts from Alpeter on working on automation projects internally vs. outsourcing them: “I would say that when it comes to outsourcing, some of your best practices are going to leak out into the industry — but you’re also gaining all the best practices of the outsourced industry. I recommend looking to that if you’re trying to catch up.
“But if you’ve got something that is a secret sauce that you don’t want anyone else to know about, then that’s where I’d think about trying to build something internally,” Alpeter concludes.
It’s time to charge your AI “trust batteries”
Of course the discussion was going to make its way here eventually.
“A lot of conversations have been had in the last year and a half about AI taking over everything. What is the practical application of AI and automation in general?” Alpeter asks Yaseen.
“I think it’s really important in every industry everywhere,” Yaseen responds. “I’m pretty optimistic that it’ll make a lot of new things possible.”
…but not without building some trust and leveling the playing field first.
When it comes to automation, Yaseen notes that most of Parabola’s customers have automated the obvious and are familiar with the landscape: “They’re not naive about all of the tools that exist, and they’ve already automated the easy to automate. The things that are left are really bespoke and complex and they change really often. As a result, there’s a lot of manual attention that’s needed, along with subjective judgment.”
This is where AI can be extremely powerful.
“I think the newer AI models have enough ability to handle dynamic, subjective, non-normalized data. They can actually take a decent amount of that logic and bake it into at least a prompt for these large language models if you are careful and do it in a way that has enough guardrails and protects against hallucination,” posits Yaseen.
But that’s exactly the challenge. Recent news like Air Canada’s customer service chatbot making up policies and dispersing them to passengers (and the lawsuit that followed) can make it feel daunting for businesses to experiment with AI, but Yaseen doesn’t think that should stop you from responsibly weaving it into the way you work.
Not only can AI be increasingly viewed as a sort of junior analyst on your team, capable of performing rote tasks and infusing some decision making and analysis here and there, you don’t have to be in the dark about how and where it’s being used.
Collaborate with AI like you might a colleague. “One of the best ways to communicate is to just show your work and make some decisions together. That way you know if you’re talking about something totally different and can hash it out, or you might see that you’re looking at reality in the same way.”
Yaseen views this as a very “trust battery charging” activity to do. You can do this with AI by understanding the process: You just need to know why the end of the pot roast was originally cut off.