Use LLMs

If you are not using LLMs extensively in your work, you are making a huge mistake. Every day you don’t, you are falling further behind your peers and competitors.

Leveling the Playing Field

I don’t know if it’s true, but disgraced (but also world-renowned) cyclist Lance Armstrong once said:

In other words, if ALL of your competitors are using something and you aren’t, you simply cannot compete. You are at a disadvantage. If you want to stay relevant, you have to use the same tools that your competitors are using.

In this case, I’m not suggesting that you take Performance Enhancing Drugs (PEDs), but with AI being where it is, I am saying that you simply MUST be onboard with using LLMs, to stay relevant. This applies to every industry, but especially within Information Technology (IT) and Software Development.

What do I need to know?

Assuming you are brand new to LLMs, there are just a few key concepts that you need to understand to get started. This is not an exhaustive list, but it is a good starting point.

What is an LLM?

A Large Language Model (LLM) is a type of artificial intelligence that is trained on vast amounts of text data to understand and generate human-like language. They are designed to perform a wide range of natural language processing tasks, such as text generation, translation, summarization, and question answering.

Training LLMs

The big deal initially was that these LLMs were trained on billions of parameters, which allowed them to understand and generate human-like language. The intitial training data was things like: all of Wikipedia, all of IMDB, all of AllMusic, all of Reddit, all of StackOverflow, etc, etc.

The first attempts took over $100 million dollars and months of time to train. Since then, there have been crazy advancements in the field - like Deepseek (mentioned above). China came up with a model that was equivalent to GPT-4, but trained in just 3 days for only $10 million dollars. This is a testament to how quickly the field is advancing.

So “training” the initial knowledge of an LLM, that is still something that Big Tech companies do. They produce these “base models”.

Using LLMs

Once there is a base model, Big Tech (or anyone else) can then “fine-tune” the model for specific tasks or industries. GitHub Copilot, for example, is a fine-tuned version of OpenAI’s GPT-4 model that is specifically designed for software development tasks. It has been trained on millions of lines of code to understand programming languages and provide relevant suggestions. That included private code from GitHub too.

You can use these Big Tech LLMs in the websites that they provide, like ChatGPT, Gemini, Deepseek, etc. You can also use them in your own applications via their APIs.

Retrieval-Augmented Generation (RAG)

Retrieval-Augmented Generation (RAG) is a technique that combines the power of LLMs with external knowledge sources, such as databases or search engines. This allows LLMs to access up-to-date information and provide more accurate and relevant responses.

The basic idea is this: you ask your question, and the RAG system (typically an API endpoint) will first use a mini search engine to find relevant data based on the keywords in your question. Then, it will pass that data, along with your original question, to the LLM. The LLM will then generate a response based on the retrieved data.

There are no turnkey RAG systems. You generally build something like this yourself. You can use existing libraries and frameworks, or there are web UI’s like N8N that let you build a RAG system without writing any code.

Model Context Protocol (MCP)

In November 2024, Anthropic announced this Model Context Protocol (MCP) as a way to standardize how LLMs handle context and memory. Specifically, it’s a standard way for a system to offer two services to an LLM:

1. /resources: This endpoint provides the LLM with a list of resources that it can use to answer questions. These resources can be anything from documents, databases, or APIs.
2. /tools: This endpoint provides the LLM with a list of tools that it can use to answer questions. These tools can be anything from search engines, calculators, or other APIs.

Also see: MCP GitHub Organization

The Players

Below is a table of all of the Big Tech players in the LLM space, and their respective offerings. This is not an exhaustive list, but it is a good starting point starting from largest market share (at the moment - in May 2025):

Prompt Engineering

Prompt engineering is the art and science of crafting effective prompts to get the best results from LLMs. It involves understanding how LLMs interpret and respond to different types of input.

The main takeaway is understanding that LLMs know everything. Like, all human knowledge: the contents of every book, every article, every movie, every song, every piece of art, all history, in all languages, all of science, all of math, all of philosophy, etc. They are trained on everything that humans have ever created. They are the sum total of human knowledge.

With that is mind, that vast knowledge MUST be focused. So, by default, most of these LLMs have a standard prompt of something like: “You are a helpful assistant.”. That means when you ask a question, it will try to answer it in the most helpful way possible. It likely won’t be specific to your needs.

Well, with Prompt Engineering, this simply the concept of telling the LLM to focus on how it answers your question, from a specific perspective. For example:

• “Act as a seasoned, principal-level software engineer who is passionate about clean code, best practices, and eliminating technical debt. Please create a function that takes a list of integers and returns the sum of all even numbers in the list.”
• “Act as a world-renowned chef who specializes in Italian cuisine. Please provide a recipe for a classic Margherita pizza.”
• “Act as a professional financial advisor with over 20 years of experience. Please explain the benefits of investing in index funds for long-term wealth building.”

This “one simple trick” can drastically change the quality of the response you get from the LLM. It makes all the difference.

Summary

If this seems a little too abstract, you might be thinking of how you can use LLMs in your day-to-day work. Here are some practical examples to get you thinking in this direction:

1. Explain something: Did someone just use an term, acronym, or jargon that you don’t know? Ask: “What does [term] mean?” or “Explain [term] to me like I’m 5 years old.” or “What is the difference between [term1] and [term2]?”
2. Get Ideas: If you explain your problem to the LLM, you can then ask it for ideas. You will be shocked at how good and novel the ideas are. For example: “I need to come up with a new feature for my app that will help users [problem]. What are some ideas?” or “I want to build a micro-SaaS. Create an table of 10 product ideas that could be built in a month, sorted by potential revenue.”
3. Dinner Ideas: Tell the LLM what you have in your fridge and ask it for dinner ideas.
4. Crash Course: At work, you just found out that you need to learn a new technology. Ask the LLM for a crash course on it. For example: “I need to learn [technology] for work. Can you give me a crash course on it?”
5. Fitness Coach: Ask the LLM to create a fitness plan for you based on your goals, current fitness level, and available equipment. For example: “I want to lose weight and build muscle. I can work out 4 times a week for 30 minutes. I have kettlebells and stretch bands. Can you create a fitness plan for me?”
6. Travel Planning: Ask the LLM to help you plan a trip. For example: “I want to visit [destination] for [number] days. Can you help me plan my itinerary, including must-see attractions, local cuisine, and cultural experiences?”. This is particularly well-suited for “Deep Research”, like OpenAI’s ChatGPT supports. It will take like 10+ minutes and deeply research your destination and give you an extremely high-quality itinerary.
7. Writing Assistance: Ask the LLM to help you write an email, blog post, or even a book. For example: “I need to write an email to my boss about [topic]. Can you help me draft it?” or “I want to write a blog post about [topic]. Can you help me outline it and provide some key points?”
8. Interview Preparation: If you have an upcoming interview, ask the LLM to help you prepare. For example: “I have an interview for a [position] at [company]. Can you help me prepare by providing common interview questions and tips on how to answer them?” Better, you can paste in the job description and upload your resume - which will give you much better results.

Hopefully this gets the wheels turning about how LLMs can give you help in all sorts of ways. You really can’t go wrong. Ironically, you can even ask the LLM to you work better with that LLM, and it will coach you through!