Inkling: Personal AI Shopper
An AI-driven personal shopper that is able to recommend the right product based on the unique needs and preferences of a buyer.
Overview
Inkling was an experiment to imagine what the intersection of generative AI, consumer reports and e-commerce would look like. The result, was a consumer-first platform that learned about the unique needs and preferences of a buyer and then provided dynamic help to purchase a desired good or service. The service was pretty magical.
After completing a personalized survey, the platform would create a buyer persona for the individual. That individual could then shop for themselves, or for another individual (gift, holidays etc..). Shopping entailed entering the general category of the product (i.e. BBQ or impact driver) which would lead to dynamically generated questions (based on the persona) to refine the recommendations. The best matches were then displayed with the rationale as to why they were selected.
This project was an incredible showcase of what is possible with the advancements of generative AI, and that companies can create amazing user experiences without needing deep deep AI expertise.
Client
Inkling was built via a joint partnership between Brad Moggach and Chris Young.
The project was an intensive effort over a number of months to explore the technology, refine the user experience and attempt to develop a sustainable monetization strategy with the eventual goal of establishing Inkling as a standalone company.
Technology
Inkling was built with the standard Harled stack to support moving fast and optimize for learning and integrating user feedback.
Gratitude
We were fortunate to quickly develop an interested community including venture capitalists, other techies and end users. Their feedback was incredibly helpful in quickling building a product that we found we couldn't live without.
We're also very thankful for those pushing the AI frontier forward. We benefited greatly from the large commercial models and services as well as open source models.