The Friction Of Selling Hermae

June 14th marked the four month anniversary of incorporating Hermae.

So much has happened since my initial tweet back in January. I never would've predicted that this small experiment would lead to demos with Puma, Siemens, Carlsberg,, Jio, C.H. Robinson, ADT, Mercury, and so many more. Crazy!

All startups are bets. Our original bet was that an intelligent assistant trained on a company's design system would drive efficiency improvements and help design system teams improve adoption across their organization.

While this may or may not be true, the business opportunity around this bet has changed in substantial ways since January. Through our conversations with prospective customers, we've been witnessing first-hand the diminishing demand of such a product.

AI Has Changed

The landscape has evolved dramatically since January. AI has become mainstream and with that, we lost our edge of providing a new "wow" experience. The impressiveness of our demo has declined over time as people have become more familiar with the technology, and with this maturation, incumbent tools like GitHub Copilot have grown in market dominance.

Our posturing as a privacy-focused alternative has weakened as teams have become more comfortable using mainstream AI products in their workflows. Similarly, building tailored experiences like the ones we offer has also been made easier with OpenAI's Assistant API. As the accessibility to the technology improved, our defensibility weakened.

Designers Not Included

Selling to design system teams is difficult. They are notoriously under-invested and the persona of the team lead is inconsistent. Sometimes it's an engineer, sometimes it's a designer, and sometimes it's a PM who doesn't understand either. With our custom implementation approach, we'd usually need to talk to an engineer to learn more about their architecture so that we may perform the training and supply our product. If the team was led by a designer, this point would instantly fall flat and navigating these conversations would instantly stifle sales progress. At the same time, the product was not built for designers, and they would leave disappointed if they were on the call. As it turns out, if the product doesn't serve everyone on the team that's looking to buy, it won't convert to a sale.

No One-Click Solution

With the difficulty of navigating those conversations, it would be amazing if we could offer a one-click solution to teams with a self-serve model. Unfortunately, all design system documentation sources are different. Some use Storybook, some use Figma, and many build their own websites. This makes ingesting the documentation for training a tricky process that can't be fully automated. So, that leads us back to the custom experience sales process. Too much friction.

Through our sales conversations, those three points have made themselves clear as the primary blockers to continuing purchasing discussions. Those three points are unchangeable. But something has to change if Hermae is going to survive: the product or the business.

AKA, time for a pivot. That's what I'm exploring now.