We are a book summary company readingraphics.com. In the last 2-3 years, we have been exploring the use of AI/ML for various parts of the business, with limited success. With the exponential growth in generative AI and other ML capabilities, weâd like to revisit those efforts.
BROAD OBJECTIVES
There are several areas we have been looking at, including:
1. Using AI/ML to replicate, simplify, or enhance our existing business processes or output quality. Examples include:
- Generating book summaries (text). Weâve already tested ChatGPT4, hugging face, etc. The quality if very far from the the details and accuracy we require, so we're only using for reference at this point to supplement human writing.
- Generating the first cut of our book summary infographics (again, our infographics are extremely detailed to capture an entire book and existing AI capabilities seem limited). Still infographics generators come close
- Generating blog, email, social media content etc. using a mix of our existing content + third party content (weâre already doing this but may not be fully leveraging the capabilities, especially the combination of text + graphics).
- Create "content nuggets" (text + graphics) using our own summaries and infographicsâthis is tied to (2) below
2. Leverage on our existing content (and potentially tap on external content) to create derivative products/services that can generate extra revenue streams and/or improve value to customers. Here are some examples of ideas (generated through chatgpt4) which we think are interesting. The question will be: which ones are the most viable and how can we prototype and test the concept quickly?
- Customized Learning Paths: Use ML to understand the learning needs and preferences of each user. Create a customized learning path that combines summaries from various books..
- Cross-Referenced Summaries: Create a system that can identify related concepts across different book summaries. This will allow users to see different perspectives and ideas about the same topic from various authors.
- Interactive Summaries: Develop an interactive AI tool where users can ask questions and the system would pull relevant information from different book summaries.
- Problem-Solving Guides: Compile a series of book summaries that are focused on solving specific problems or achieving specific goals, like "how to start a new business" or "how to learn faster".
- Personalized Recommendations: Use AI algorithms to understand users' reading habits and preferences, then recommend them relevant summaries. This could be based on their reading history, the ratings they've given to previously read summaries, or their interaction with the content.
- Trend Analysis and Predictive Summaries: Analyze reading trends among your users to anticipate what topics or books they might be interested in the future. Provide them with summaries of upcoming books or trending topics.
- AI-powered Flashcards: Create AI-generated flashcards for each summary to reinforce the key takeaways. These flashcards could adapt to each user's understanding and recall ability, focusing on concepts they struggle with and repeating them at optimal intervals for efficient learning.
- Deep-Dive Summaries: Use AI to create 'deep-dive' sections in your summaries, pulling in extra information on a topic from multiple books, and summarizing it further. This will provide users who want more in-depth knowledge with the extra detail they're looking for.
3. Other forms of web experience personalization and enhancement (ranging from gathering feedback to providing on-demand answers to product recommendations). We tried a few existing vendors (e.g. Pure Clarity and Personyze) a few years back but the solutions weren't that great. Not sure if there are better solutions out there for our business.
4. Identify book categories and keywords for each title to help us with tagging (UX + content management purposes) + SEO. We started using ChatGPT 4 to develop a first cut, but not sure if there are much better ways to it (e.g. auto-tagging?)
PROJECT SCOPE
Suggest to move in 2 phases
1. Obviously we canât go into everything above. So we will probably start with a consultancy to lay out the considerations, strategy and approach. Possibly identify a few low-lying fruits that can be achieved easily with our existing content and current AI/ML capabilities. Then map out a prioritized plan of sorts. [Suggest a budget of $250-500? This is an arbitrary number as we have no idea whatâs involved here. Please feel free to counter-propose (but kindly explain whatâs included in the scope)]
2. Move into execution, including concept development, prototyping, user feedback etc. We have a really lean team, so ideally the bulk of the process can be handled by AI to keep things manageable. [The budget for this will depend a lot on whatâs decided in (1), we can confirm this later.