Meet EJO -Aftermarket Parts Assistant Bot
An Underserved Industry and An Old School Chat Bot
My father worked for decades for OMC before it was bought by BRP™ and, in 2020, BRP announced that they were not going to continue manufacturing the outboard engines that built my father's career as a field technician - Evinrude-Johnson™ outboards, making one less American-based manufacturer to compete against the growing number of Chinese owned and Japanese owned outboard engine manufacturers. When parts production was announced to continue and the feedback I heard from real dealerships I had been working with (See Beck's case study, St. Augustine Marina), I knew that there would be a communication bottleneck that would trickled down to impact all dealerships that continue to service and repair Evinrude-Johnson engines once manufacturer support is cut off completely.
That is when I thought about designing a chat bot assistant for dealerships and mechanics that still offer repair and service for Evinrude-Johnson™ models that were purchased previous to 2020. These models can only be referenced from the manufacturer catalogues for that year, many of which are only available in print form.
When my father and I discussed selling parts out of his garage, I realized that parts matching would be the hardest part of this project. That was when I decided that I wanted to move forward with EJO and I started by training Anthropic's new Fable modal on the literature, building the knowledge base for what you can now interact with in this demo.
EJO was consciously decided to NOT be powered by LLM APIs. This was done because it will confidently give users incorrect information on an inconsistent basis. While AI was used to make EJO the amazing bot that it is, it is backed by pure code for both efficiency and resource reasons. As it is, EJO can easily be deployed on virtually any server in any database and integrates well into most websites without changes to the framework behind them.
Changes I plan to make to EJO in the future incldue support for other manufacturer catalogues and broad support for multiple models, engine specs, and perhaps other industries.
Created by Samantha L. - Last Updated July 2026
- EJO can be branded and is white label, can be deployed on virtually any dealership website or used as a standalone application.
- Designed using AI trained on manufacturer catalogues and print data not currently transcribed into a digital framework. EJO will be updated to adapt to other manufacturer's catalogues and can work as a customer service bot or directly with technicians to properly identify and order parts.
- EJO was consciously designed without AI APIs in order to preserve the number one goal - accuracy. Current LLMs will often claim a part is correct, regardless of specific details like transom or shaft measurements that directly effect the overall build out and incorrect parts orders hurt small businesses / dealerships.
- EJO identifies with old-school chat bots but its improved understanding of user input directly shows its LLM powered build out, allowing for users to interact with the bot efficiently regardless of language use, misspelling, etc.
- As LLMs improve and access to advanced models becomes more open, EJO can be expanded to act as a companion bot or assistant out in the field to solve problems, handle customer complaints, diagnostics, and more.