The Use of AI at Diamant

At the end of the day, Diamant is a bag company.

Our days are spent analyzing ad performance, choosing new bag colors, and debating products to build.

We do not ship code, develop LLM models, or get close to all the cool work the tech industry is doing right now. However, I think we can all agree that the AI world is crazy right now, as it has clobbered all of our Twitter (or X) feeds.

Every week there is something new coming out, whether it is a new model, a new feature, or a fix to an old version — it makes it seem like we are on the clock to take all of these new tools and use them because "you are falling behind if not."

In a way, this is true. But when your day-to-day is backpacks, keeping up with progress in AI looks a lot different than what Twitter makes it out to be.

Three months ago I made the jump with Diamant as my full time job. Unlike Gavin and Shai, my background is in technology and data analysis. My job at Diamant is simple: make sure sh*t goes smoothly. Whether that is budgets, recruiting, distribution, margins, finances, retail…. Just make things go smoothly.

While AI may not take over choosing the right color for a Weekend Warrior, I figured it could actually make a really solid impact on the company’s day-to-day. As any product manager (yes, my old role back in corporate America) does, the first step was to be curious and understand what were some of our key problems to solve.

We spent days trying to identify which of our processes could be automated. In our world, a majority of the work is manual, yet some tasks tend to be repetitive in nature. For example, when we receive a sample from the factory, we need a person to actually look at the bag and test it. Not many repetitive steps here, plus every bag is different.

However, when it comes to processing returns, not only is the process manual, it is also incredibly repetitive. A customer emails, we check if they meet policy, request a label from our warehouse, send them an email with the label, and on and on.

We were spending 2–3 hours a day during peak season on Customer Support. We needed to buy that time back.

So we now have a problem that we know is manual and repetitive on one hand, while also having AI and automation on the other. What next?

Well, my approach first started by asking: how can I break down this one process into sub-processes that I can then automate one at a time to create a singular cohesive loop? Wow, that was corporate, how about: how do I simplify all of this? 

After discussions with Shai (who was bravely managing this process and replying to your emails by hand), we were able to define our first step: the yellow pills. Requesting the label from our 3PL and then automatically sending it to the customer was the first piece. With the help of Google Scripts, Claude, and some coding background, I was able to isolate this return label loop and automate it.

Now, as soon as we approve a return, we automatically notify the warehouse, they send us a label, and our Gmail automatically sends that label to the customer.

Boom: Automation and many man hours saved.

Eventually, we were able to build out an automated process for every pill in that graph, step by step. 

This is just one small example of how this boot bag company can take that next step and tap into the AI boom that everyone talks about. We definitely don't fully understand it yet  — but here we are, with something kind of cool. 

We will continue to identify other issues that come our way, and see where AI can help. Another area we're developing is a way to support questions about the products themselves. Think "will my boot fit in the 40L?". I think our AI friend here could help with that as well. The key here is our identification of processes rather than a burning desire to just "implement ai into our business."

AI is helpful, but utilizing it day to day has to be methodical. Unless you find problems that you need solved, you may end up wasting more time asking Claude questions than by just doing what needs to be done.

It's that balance that's the struggle of it all. Because the AI rabbit hole is real and if you don't give yourself some sense of direction you'll drown. At the end of the day, that's one of the start up struggles. No one said going from 1 to 2 would be easy, but hey, it is more fun like this anyway.

-e