Case studies

Results

What it actually looks like to work together: the problem, what we built, and what changed.

In their words

“It was way less daunting than I expected. You did such a great job understanding and synthesizing my problems, laying them out and coming up with solutions.”

Jenna, Powder Room Beauty

Powder Room Beauty

Late-night inquiry emails became a quick review and two clicks.

The situation

Jenna runs Powder Room Beauty, a makeup education business offering private lessons, group parties, and pop-ups. She's constantly on the go and rarely at her desk, which meant inquiry emails were piling up and getting answered late at night. The replies were mostly the same questions over and over, but each one still required her full attention to draft and send.

What we built

We started with a Claude 101 session to get oriented, then built an inquiry responder agent. It reads new inquiry emails, figures out what type of inquiry it is, and drafts a reply in Jenna's actual voice using type-specific templates she maintains in Notion. It answers extra questions using her website and a Notion brain doc.

Anything it wasn't confident about gets flagged in red at the top of the draft for her review. This is one click to run when inquiries build up, and sending becomes as easy as reviewing & hitting a button!

Refound Goods

Hours back every week on delivery admin.

The situation

Refound Goods is a curated secondhand furniture marketplace in Denver. Every order needs a human to agree on a delivery time, coordinate with a driver, and send enough notifications to make sure someone is home to receive it. Before we started, Katie was emailing customers one at a time, multiple times a week, and hand-writing the weekly schedule for their delivery partner. It worked, but it ate hours every week and left a lot of room for things to slip.

What we built

We rebuilt the delivery workflow end to end. It started with a revised intake form that gave us clean, structured data to work with. From there we built a Claude skill to transform that data in one click into a format that could be uploaded into OptimoRoute, a route planning tool Refound was already paying for but not fully using. We set up automated customer notifications via email and SMS, proof of delivery, a driver-side app, and a Claude skill to generate the daily packing instructions PDF for their driver.

We also made pointed decisions about where AI would be overkill. The important part of this was not 100% automation, but creating a system that took work off Katie's plate in a trustworthy fashion.

Results

In their words

Powder Room Beauty

Refound Goods

Have a problem worth talking through?