📖 Know Your Benefits (Idea Edition #8)

🧠 Top of Mind Technology

• Sick and tired of hearing about AI but not knowing how it’s implemented? Check out this article which highlights not only what a Large Language Model (LLM) is, but how one is built (with an example).

• The future isn’t all software and AI. Tesla is changing the game by using a manufacturing method called Air Bending to better shape and preserve stainless steel.

• Our favorite YouTube channel for engineers right now is Stuff Made Here. Shane, the owner and creator, has built some of the coolest devices including:

  •  A basketball hoop that doesn’t allow you to miss.

  • A haircut giving robot.

  • A giant wall painting robot.

🍪 Business Bites

• With its $3 Trillion dollar valuation, Nvidia is worth more than $100 million per employee.

• If you’re thinking about how you’ll be able to afford a home one day in this crazy market, read up on a similar market from the 1980s.

• Ramaco Resources ($METC), a small-cap coal-mining company, could be sitting on a literal gold mine. The $2 million mine they bought in Wyoming could one day be worth $37 billion.

Idea Feature #8: Understand your benefits and resources like never before

⚾ The Elevator Pitch

Leverage an existing LLM chatbot like ChatGPT to build a benefits and resources analyzer and guide for insurance plans, employee benefit plans, health care, and more.

💡 The Idea

Given the thousands of dollars Americans spend every year across different types of insurances (health care, auto, home, life, etc.), it would behoove each person to know more about what benefits they get as part of their policies. For instance, what if your auto insurance plan offers free 24/7 towing? What if your health care plan offers monthly massages? Perhaps more importantly, what if you could get more coverage for cheaper with a different provider? Even if you think you know all of the policies and benefits you receive, double-checking doesn’t hurt. We want to build a tool that reads through the official documents of each plan and insurance offering a person has, and output a 1-2 page document that highlights the key benefits and costs of each.

👷 The Work

1. Don’t build your own LLM. ChatGPT or another large-scale model like Perplexity can do that for you. You can also use their API and essentially build on top of the infrastructure they have created.

2. Test if it will work - Take your own health care insurance, employee benefits, auto insurance, and additional insurance plans and feed them to ChatGPT or whatever LLM you choose to use. Confirm that the LLM can tell you what your benefits are through a manual review of your own plans’ documents.

3. Create a landing page that will allow a user to upload multiple documents (from each of their plans).

4. Create a pdf generating software to take the output/benefits from the LLM and place it nicely onto a 1-2 PDF.

5. Synthesize the process so the landing page, LLM, and PDF software are all in sync and can provide the output the user requests.

6. Improve the offering over time by taking the following steps:

   1. Add in a guide to help users find where their official plan documents would be located on their provider’s website.

   2. Expand beyond insurance summaries and prompt users to add the policies and benefit documents of their credit cards, banks, and other financial institutions.

   3. Based on the current coverage a person has, recommend providers with similar coverage that offer a cheaper rates if able.

💸 Revenue

• Charge a small one-time fee for use of the tool - We would also recommend using a freemium model. When the output report is generated, give the users part of the summary for free, but make them pay the fee to access the report in full and to download it.

• Potential Commissions – If you add the ability to recommend other insurance policies/plans, you can generate commissions from those insurance providers for providing leads and eventual sales.

🔻 Expenses

• Hosting and website design fees

• LLM API usage costs

• Advertising and marketing

👍 The Good Stuff

• Leveraging Great Tools – ChatGPT and other LLMs are already great at reading and synthesizing documents. Using them is highly beneficial as the most complex part of your tech stack is already developed.

• Low Expenses – Presumably the number of API calls will be very low, and the additional website expenses are very minimal.

👎 The Risks

• Data Safety/Integrity – Many insurance policies contain Personal Identifiable Information (PII) and other sensitive information. Users may be concerned about uploading that information to a website they have never heard of, and most would want to ensure their data is not being stored or sold.

• Too much user effort – It may take users a substantial amount time to track down all of their official policy documents, which is something that could derail their eventual usage/purchase of the product. This is also why we recommend creating guides to help users track down their documents.

🤔 Musing of the Week

What will happen to SEO and websites if AI tools such as ChatGPT becomes more prevalent?

SEO (Search-Engine Optimization) has been an invaluable tool for internet creators and businesses for years. SEO is essentially a methodology to help you rank higher on Google and other search engines. However, the long-term future of search engines is somewhat unknown with chatbots becoming more and more popular. The most popular chatbot, ChatGPT, has had over 180 million users less than 2 years since its public launch.  

Let’s assume for a second there is a world where you open up your search engine and are presented with a ChatGPT style chatbot instead of the typical search page. You ask the chatbot a question: What are some good places to eat in New York City? You are then presented the following results:

In this hypothetical world, as a restaurant owner, how could you become one of the restaurants listed in the ChatGPT output? On Google, you can currently leverage SEO to be one of the top search results on the internet for key searches/words. Perhaps for ChatGPT/Chatbots, you will need to become the most prevalent part of the dataset the chatbot uses. Per Human SEO, the emergence of chatbots will force SEO experts to transition to “define the site’s foundations to be properly crawled and indexed,” and “will need to focus on working as a bridge and translator for AI and robots to understand content correctly and more efficiently.”

📚 For the Road

• The best ChatGPT alternatives

• There Is Platinum in the Dirt on the Side of the Road