Dynamic Solve
Making solve more intelligent
Last updated
Making solve more intelligent
Last updated
When the first look of consumr.ai was worked upon, it was a long list of fields filled with validations. We knew that it was really tough for a user to comprehend with such large forms. Hence, the thought to build an intelligent sparked up in a room in San Fran and a month later, Omnibox was born. However, over a period of time we realized that for some users to select the right inputs was stressful and entering the wrong input will give you results that are way off.
Then came a series of thoughts and experiments with the likes of Audience Navigator. But after a few failed attempts and rolling back some developments, we knew that recipes also known as solve is the way to go.
Taking inspiration from MS Word and how they showcase templates, we built the solve section, with a hope that we will add 100+ recipes to it and build a marketplace of use cases. Call it wishful thinking, it didn't happen the way we would have liked and Solve was reduced to a small stack of 8 cards instead of an exciting marketplace of solutions.
In the meantime, our access and exploration with Generative AI was becoming stronger and more promising and we started dreaming of a day when we can create a series of prompts that can create dynamic recipes based on a simple question asked by a user. This question should be a business problem and the Dynamic solve will give you a series of reports to execute sorted as per degree of recommendation.
Nothing changes. But things will be added on. There will be a parallel screen, when you go to Solve. The default screen is the same static solve screen with a few static cards. The fun is when you hit the button, "Ask me a question".
You will see a search bar where you will be required to enter a business problem in the form of a question. For example: How does my brand compare to {name_of_a_competitor_brand}?
consumr.ai will start writing up each card by dynamically coming up with a curated headline, then the use case and an explanation. One card after another and will stop at 3 cards. If you want more recommendations, click the "get more recommendations" link at the bottom of the page and it will build you 3 more.