 | Episode Cheat Sheet |
EP656
24 AUGUST 2026
54 MIN LISTEN
Justin Bausch & Felisha Mina
Founder & CEO, and Co-Founder, Style My Wall & InHaus Digital Agency
| Brought to you by |   |
THE TL;DR
Justin Bausch was selling art on Facebook Marketplace as a side hustle while working in CRM at Ryderwear. Then he went all in.
He bought a printer, hired his nephew, built a Shopify store and took Style My Wall to seven figures in its first year, growing from one staff member to 22 inside two years, all print-on-demand.
Frustrated that no single tool showed his actual daily profitability, he spent 11 months building his own AI platform in Claude Code, which now runs his ads on autopilot.
With his partner Felisha Mina driving founder-led content, a revamped LinkedIn brought Mattel knocking for a licensing deal.
| Watch the Full Episode |
| From Bushy |
When a customer sends through a photo of a broken frame, nobody at Style My Wall looks at it. The AI reads the photo, determines the frame can't be repaired, spins up a returns portal, books the courier, raises the replacement order in Shopify, and tags the dispatch team. Justin Bausch built that himself in Claude Code.
Last time Justin was on the show, back in May 2024, he was at Ryderwear talking CRM, and I remember these framed prints kept creeping into the corners of the shot. "That's my little side business," he said. That side business is now seven figures, 22 staff and two Melbourne warehouses, built alongside his partner in life and business Felisha Mina.
I've been sceptical of homemade software running the core parts of a business, but this is a great example of a next-generation ecommerce business built around AI.
CHEERS, BUSHY
Tap any time code anywhere in this cheat sheet and the episode opens on YouTube at that exact moment.
From Ryderwear side hustle to a seven-figure first year04:55 Why selling art online has almost no major competitor06:44 Founder-led content that brought Mattel knocking08:43 The hidden trap of scaling print-on-demand14:48 Why he built his own profitability tool in Claude17:27 Putting ads on autopilot, with guardrails21:29 When the AI went rogue on customer emails35:55 The Bali dispatch fix that updates Shopify and Klaviyo itself36:44"I'm not some random AI guru from his basement. I've created a tool that works for me."
— JUSTIN BAUSCH | Style My Wall
| 02 The Big Lesson |
Give AI the whole process, but make it earn the keys
Justin spent 11 months building his own AI platform in Claude Code. But it wasn’t just building. There was a clear order. Nothing got the keys on day one. Here's how he staged it.
RULE 01 18:03
Connect profit, not just performance
Google Ads shows ROAS, but it doesn't know last night's net profit after staff wages. Justin plugged Xero, wages and expenses into his platform so it judges every ad against real profitability, not vanity ROAS.
Google Ads doesn't know how much you made the day before, your net profit after everything.
RULE 02 21:52
Earn autopilot, don't switch it on day one
For three or four weeks, the system only recommends. Once its calls prove accurate against net profit, you toggle autopilot, and it pings you in Slack every time it moves budget, with a one-click revert.
It gives recommendations every day on where you should scale and where you should cut back, to the point where it was identifying areas that I couldn't.
RULE 03 37:02
Once it has earned it, give it the whole chain
Stuck in Bali during a big sale with dispatch blowing out to five weeks, Justin had the system track the real gap between an order going in and landing on a doorstep. The second that gap grows it updates the shipping window on the site, the Klaviyo flow and the tracking page on its own.
We need to over communicate, and that's something now that the app does. I don't have to look at it any more.
Once it is all firing, it becomes a tested and trusted sequence. The power is not in a library of spot tasks that AI can take over but a chain of events that can be gradually rolled out to take over key business processes. And we’re only at the start of what’s possible. Fascinating times.
3-4 Weeks
The amount of time Justin made the AI prove itself before it was allowed to act on anything.
| 03 Steal These Tactics |
1
Plug your real costs (wages, expenses via Xero) into your reporting so AI judges ads on net profit, not just ROAS. (19:15)
2
Audit your branded search spend. If nobody else is bidding on your brand name, you are paying Google for clicks you already had. (46:20)
3
Revamp the founder's LinkedIn and post owner-led content; it's what put Style My Wall in front of Mattel. (09:20)
4
If you spend under ~$4k/day on ads, question whether you need a paid media agency at all. (28:14)
5
Ask AI the simple segmentation question: export your Klaviyo list, pull the business owners, and build a wholesale pipeline from customers you already have. (42:10)
| 04 The Stack |
The server the platform runs on, pulling and refreshing his Shopify, ads and finance data every 30 minutes so the numbers are always live.
Justin built the entire InHaus platform in it over about 11 months, testing each piece against the paid subscription it was replacing before he cancelled anything.
CRM and flows. The platform reads his segments and product margins to recommend offers, and auto-updates post-purchase emails when shipping windows change.
Customer service inbox. The AI triages tickets and drafts or, once trusted, sends responses, including reading photos of damaged frames and deciding whether they can be repaired.
Plugged in for wages and expenses so the system can calculate true daily net profit, LTV to CAC and AOV to CAC.
| 05 If This One Landed |
EP437
Post-Purchase Potential: Turning Customer Service into Growth with Hamish McKay
Six hundred support tickets on every drop day is the kind of problem that either buys you software or makes you build it. This is the operator who built the fix himself, then turned it into the business.
EP411
Chris Clark from Milton & King: Reinventing Wallpaper For The Digital Age
Print on demand solves your stock problem and hands you a quality control problem instead. Here's the wallpaper version of that trade-off, from a founder who pivoted after COVID and the Brisbane floods and scaled into the US.
EP578
Stop Customer Problems Before They Start: Inside Keeyu's AI Ops
Telling a customer their order is running late before they have to ask is most of the battle. This one goes a step further, catching the lost parcel or the failed payment before the customer ever sees it.
| Pass It On |
Episode Cheat Sheets are built to be forwarded. If someone in your world is scaling a DTC brand or eyeing an exit, send them this one. If a colleague sent you this, subscribe for future Cheat Sheets here.
| Continue the Conversation |
Would you hand your ad budget to something you built yourself, and how long would it have to prove itself in draft mode first? Plenty of operators in here are somewhere on that curve already. Come and argue about where the line sits.
Free to join. 700+ Australian ecommerce operators talking shop every week.
