shepx reimagined: a brand experience lab

A Quick History Lesson

In 2018, we set out to create a spinoff of Brand Shepherd focused on just two services: websites and logos.

No brand strategy.

No brand architecture.

No positioning work.

Just a clean, focused option for brands that needed the tangible deliverables without the full strategic engagement.

That idea took four years to fully come to life. Between 2018 and 2022, plenty happened, including a global pandemic and a relocation of the Brand Shepherd offices.

But by 2022, the spinoff had launched.

It was called shepx.

It served the customers who wanted exactly what it offered: a logo, a website, and nothing else.

shepx has been a good business. But in 2026, looking at the state of brand building, the signals are pointing in a different direction.

The Decision

Two things have become clear this year.

First, the client roster shepx serves fits naturally back inside Brand Shepherd. Websites have been Brand Shepherd’s number one service since 2020, and moving the shepx client base and staff into Brand Shepherd strengthens the core service line rather than fragmenting it.

Brand Shepherd’s DNA has always been for the complex, for the multi-service engagement, for the strategy-plus-execution integration that shows up in expert generalist work. That is where the value is highest.

Second, we need a space to experiment. Publicly. Without putting Brand Shepherd’s main stage at risk. A place where we can test the boundaries of what brand experience can become, show our work as we go, and not be afraid to fail quickly and openly.

So here is what is happening. shepx is being repurposed. The existing shepx client roster and staff are moving into Brand Shepherd. shepx is becoming a brand experience experiment lab.

What shepx Is For

shepx is where we take risks, try new things, and share our findings, all in service of expanding what brand experience can be.

That means public experimentation. It means inviting current Brand Shepherd clients to participate in experiments as an aside, at no additional charge, when the experiment fits their brand. It means running our own experiments without a client attached. And it means failing openly when experiments do not work, and sharing what we learned when they do.

The point is not to abandon what Brand Shepherd does well. The point is to expand the lane. We are staying firmly in brand building. We are just widening what that means.

At launch, shepx will feature four experiments.

Experiment 1  //  Web Design and Development Without WordPress

The first experiment is shepx itself.

The new shepx website has been built using a combination of Brand Shepherd’s own design work and a series of sessions with Claude, Claude Code, and Claude Design. All final messaging and copy is human-generated. AI was used to speed up and amplify the process, not to originate the ideas.

The bigger experiment is the platform. Rather than building on WordPress, which has been Brand Shepherd’s default for years, we are building on EmDash, the new open-source CMS Cloudflare launched in April 2026. EmDash is written in TypeScript, built on Astro 6.0, and positioned as a spiritual successor to WordPress, designed from the ground up for the AI era and agentic web development. It is still in developer beta (v0.1.0 as of this writing), and it has been an eager collaboration with one of our developers to see how far we can take a live brand site on it.

The verdict so far is positive. Cloudflare made EmDash familiar enough for WordPress veterans to get productive fast, but built with under-the-hood capabilities (sandboxed plugins, built-in MCP server, agent-native design) that make it genuinely suited to how brand websites will be built and maintained in the next few years. Whether EmDash becomes a long-term Brand Shepherd offering will depend on how this experiment plays out over the coming months.

Experiment 2  //  Brand Music

There is an old joke in creative services that most designers are failed musicians who needed to make money somehow.

It is closer to true than it should be. Most of the Brand Shepherd creative team plays or writes music, and I have been playing since the mid-1980s.

What has become possible in the last year is the ability to take a human-created musical idea (a melody, a demo, a rough voice memo hum) and use AI platforms like Suno to expand that idea into a fully produced, brand-specific track.

This is a different use of AI in music than prompting a song into existence, which is where most of the legitimate criticism of AI-generated music lands.

This is closer to how professional musicians already work with producers, session players, and sound engineers. The human creates the core idea. The AI executes on it based on specific, informed direction.

For brand work, this opens up something that has been quietly frustrating for years: the reliance on stock music for brand videos. Stock music is available to everyone, which means it is on-brand for nobody. shepx will experiment with producing custom brand music using this approach, tailored specifically to a brand’s vibe, and available exclusively for that brand’s use.

Experiment 3  //  3D-Printed Branded Hard Goods

This one is intentionally niche.

One current experiment involves designing a custom scoop for a hard goods client whose product currently requires customers to improvise their own scooping method out of whatever is available. Why not 3D-print a branded scoop, iterate on the design, and eventually include it in the product packaging or offer it as a separate purchase?

shepx is not moving into product design. This is not that. This is about small, delightful ways a branded physical object can improve the customer’s experience with a brand. Rapid prototyping and on-brand 3D printing are cheap, fast, and increasingly capable, and they open up a category of brand touchpoints most brands have not considered because the tooling was previously too expensive or too slow.

If these experiments produce something worth scaling, we will move to a manufacturing run. If they do not, we will learn something useful and share what we learned.

Experiment 4  //  Small Language Models for Brands

This may be the most consequential of the four, though it looks the least flashy.

Brand Shepherd has been doing AI work for brands for over a year, primarily leveraging the large public language models (Claude, ChatGPT, Gemini). What we are now developing is the next step: small language models trained specifically on individual brands, seeded with the Brand Experience Guide produced through the Vibe, Tribe & Why® process, and layered with additional brand-specific information.

The difference between a large public model producing content for your brand and a small model trained specifically on your brand is significant. The public model gives you an amalgam of everything it has ever seen. The brand-trained small model gives you your brand’s voice, your brand’s positioning, and your brand’s audience, from the first draft.

shepx will show side-by-side before-and-after examples of AI outputs (copy first, images and possibly video later) produced by generic large language models versus brand-trained small language models. NDAs prevent us from naming the brands involved, but the demonstrations themselves will make the difference visible.

Of the four experiments, this is the one most likely to graduate the fastest into a full Brand Shepherd service offering.

Where to Follow Along

The new shepx will launch soon with a single-page site featuring these four experiments front and center. An expanded site with a full feed of ongoing experiments will follow.

To keep up with shepx, follow Brand Shepherd on LinkedIn or check brandshepherd.com. Announcements will go out through both channels as each experiment matures.

The next 20 years of Brand Shepherd will look different than the last 20. shepx is one of the ways we are figuring out how.

Brand Shepherd guides brands to clarity and helps them thrive. New episodes of Field Notes drop weekly.