We are building Winks as lean as possible.

Tiny team
1% OpEx
24hr funnels

🤓 Cody Plofker reveals 5 AI tools we’re using to make it happen

🤩 Curtis Matsko invites your loyalty program onto TikTok Shop

🤑 Sean Frank shares nine ways to defend your margins in Q4

Curtis Matsko

CEO, Portland Leather

Customers Can Now Spend Loyalty Points on TikTok Shop

At Portland Leather, our customers love their bags. And we love giving them reasons to buy another one!

Our loyalty program is one example. Customers earn points when they shop, then cash them in for discounts.

People say loyalty programs are outdated. Our customers missed that memo.

We used to be on Yotpo. Now we use Rivo.

In just six months:

250K+ new members joined
53.3% points were redeemed
17.4% of sales were from loyalty

Now Rivo is bringing loyalty to TikTok Shop as an inaugural integration partner.

Starting October 28, customers can join your program, earn points on purchases, and redeem them for coupons at checkout …

All directly on TikTok Shop!

Existing members link their accounts and keep one loyalty profile across both channels. You will see their TikTok orders alongside their DTC purchases.

Rivo is launching with select brands ahead of BFCM. Get your name on the early access list!

Request Early Access

Cody Plofker

Jones Road, Advisor

How We’re Using AI to Build Winks With a Tiny Team

When we launched Winks over a month ago, I spent two hours just uploading ads.

I forgot how tedious + painful that is.

Now we are using AI to:

Upload ads
Write hooks
Create statics
Animate videos
Build landers
Build email flows
Reply to customers

And a bunch more.

Our goal since the beginning was to build an AI-first brand with a tiny team and 1% OpEx.

I am as AI-pilled as it gets.

Here’s the tools we’re using.

1. HQ: Give AI Context

I don’t want to re-explain our brand every time I ask AI to do something.

We use HQ to store knowledge + share it across our tools.

Product information
Ingredient studies
Customer profiles
Previous test results
Brand guidelines

Whenever we ask Codex to build a new lander or Claude to write some copy, it already has the most up-to-date information.

We also created dedicated skills for jobs like building landing pages, designing static ads, and making animations.

That allows us to reuse the same process + design rules without explaining each step every time.

When it gets something wrong, we update the instructions.

Both Matt and I use this stuff. If I improve a skill, he has access to it. We aren’t building separate versions of the same thing.

We’re constantly refining.

I finally understand what HQ does.

2. Runneth: Create + Upload Ads

One workflow that is becoming pretty common for me:

See an ad I like
Drop it into Slack
Ask Runneth to recreate

It has access to our HQ. So it has all the same context + assets already available.

Because of the skill we’ve built inside HQ, it also knows I want the 4:5 version first, a benefit-driven headline, proper spacing, and to leave ad uploads paused for my review.

Tbh, V1 of ads usually need tweaking. But it’s getting better. I give it feedback and ask it to save my preferences.

We’re also using Higgsfield + OpenAI’s API for image generation.

For animations, Matt uses Claude Code with skills saved in HQ so we can reuse the process + keep characters consistent.

AI is awesome for creator content, too.

We found five creators with Trybe and put them on retainer. They produce 60 videos per month. Runneth handles a bunch of the work after that.

Pulls in the videos
Decides what it’s about
Writes relevant ad copy
Picks the best lander
Uses ad naming conventions
Uploads it to Meta for review

While I was at the US Open a few weeks ago, new Trybe videos came in, I reviewed them, and told Runneth to build the ads.

They were ready for my review when I got back home.

To pull this off, we had to build a database of our landing pages, connect Runneth to Trybe + HQ, and give it clear instructions.

Worth it.

3. Codex: Build Website + Landers

We launched with seven landing pages.

Not saying they were all great. But early on at Jones Road, getting a landing page built by an agency cost like $20k.

Now Matt and I are doing the design + dev.

Primarily in Codex for the website, with some Claude in the mix.

We have a page-builder skill that analyzes any landing page we throw at it, identifies the principles that make it work, and then builds a version for Winks.

We preview the changes in Shopify and approve them before they go live.

Matt and I also built an internal website using Vercel we call Winx OS (that typo is intentional). It lets us save brand assets, ads, and design rules all in one place.

Fonts and logos
Product images
Buttons + CTAs
Shopify Components

Then I connected it to all our AI tools through an MCP, so Codex can reference assets directly when building.

This is what I always dreamed of being able to do when I was at Jones Road. Matt and I can have an idea in the morning and execute it by the afternoon.

4. Omnisend: Create Email Flows

Most of our email building happens in Claude or Codex, using the context we’ve saved in HQ. Then we use Omnisend’s MCP to put the emails into our account.

We still review everything before sending; however, we barely need to log into the platform to create broadcasts or entire flows.

5. Richpanel: Customer Service

If we had started Winks three years ago, one of the most time consuming jobs would have been customer support.

We launched with Richpanel’s built-in AI.

We pay super close attention because we need to know what our customers think. But as we grow it’ll be a huge benefit for helping us scale without adding overhead.

Could we build this software on our own?

Idk and I don’t care.

Richpanel works amazing. And I’d rather pay for a tool built by people who know that job better than we do.

We have plenty of other work to do.

Our 10/80/10 Rule

Surprise: We are still working with humans.

Our first hire was a designer/editor. We’re working with Trybe creators to make real content. We are leaning hard into partnership ads right now.

Matt and I still set the direction + everything before it goes live.

Why?

Taste still matters.

A good designer can get more out of these tools than I can. They know what to ask for and what needs fixing.

We think of our approach as 10/80/10.

10% idea + direction
80% AI does the work
10% review + fix

Just to be clear … I’m still busy af.

That last 10% can be painful.

Getting a landing page most of the way there is one thing. Getting the spacing, copy, and design exactly right can take a bunch of revisions.

Ultimately, we need enough volume to learn what works with enough quality to give it a chance.

AI lets our tiny team do more of both.

Sean Frank

CEO, Ridge

9 ways to defend your margins during q4

Everyone is about to have their biggest sales day ever.

Some of you will lose money.

Discounts, ad costs, shipping, returns.

Record revenue.
Nothing left to keep.

Levanta, the creator affiliate platform behind 3,000+ ecom brands, put together nine plays to protect your profit this holiday season.

That is more important than pumping revenue.

It has tactics from ecom experts at OMG Commerce, Redo, and more.

My takeaways:

- Stop advertising high-return products
- Give store credit before returns arrive
- Let customers exchange cross-catalog
- Do not discount products that will sell
- Shift your spend before going OOS

That is a sample.