What LLMs Are REALLY Good At

A must-have in your toolkit for getting things done

An Excellent Tool for Work and Life

I assume is safe to say that most (if not all) people reading this article use Large Language Models (LLMs) in some way, shape, or form. LLMs has already changed how we work, think, and create. As this is not the first time I write about LLMs, I previously discussed some of their limitations here on Substack. In another article, I make the case that LLMs are great assistants but terrible teachers. But today, I want to shine a light on what LLMs do brilliantly. From consuming information to generating ideas to automating the mundane, LLMs are like a super-smart and super powerful helper who’s always ready to take action.

Consuming Information (and Is It Knowledge?)

LLMs are phenomenal at helping you consume information. It is no exaggeration to think of them as a type of “data juggernaut” capable of summarizing a 300-page report in a few minutes. But is consuming information the same as gaining knowledge? Not quite. As Luciano Floridi explains in Information: A Very Short Introduction, information consists merely of raw data or facts, whereas knowledge emerges when these facts are interwoven with our personal experience and understanding. In other words, while LLMs supply you with the essential raw material, it remains your task to craft that material into something meaningful.

I wrestled with choosing an example to illustrate this powerful duo, LLMs and us, for consuming information and acting on it, where the roles of each part are well-defined, working perfectly together. I didn't want to use the “vacation planning” example since I read it so many times in other pieces and I feared that is already worn out. But thinking again about, it is actually a great example.

Imagine you’re planning a family vacation to Italy. You ask an LLM, “What are the top 10 activities for someone visiting Tuscany for the first time this summer?” The LLM equipped with online search will “scan the web” and deliver a carefully crafted list with dates, locations, and highlights. In an amazing display of information consumption power, the LLM saved you hours of Googling, so you can focus on the fun part and dream up your itinerary. Now you go home and talk to your family. As you share the results and go over the options, you notice facial expressions, reactions and comments to some details, moments of silence or a sudden enthusiastic response to a particular aspect of the plan. You remember previous experiences, what worked well, what didn't go as planned, and what you wished you knew before planning vacation trips.

The vastly detailed chunk of data the LLM put together is information. Then, based on the perception of your family's vibe, their desires, aspirations, inclinations, and preferences, you make a decision on what the plan must include. That’s knowledge

The roles are crystal clear. Looking at this example, it is evident we cannot consume information as efficiently as LLMs and LLMs cannot do our work either.

Do LLMs “Know” Anything?

LLMs don’t know anything in the way humans do. Knowledge, for humans, involves understanding, context, experience, and often an ability to reason about or apply information meaningfully. LLMs, on the other hand, are sophisticated pattern-matching systems. They’ve been trained on vast datasets of text (information) and generate responses based on statistical associations, not because they comprehend or hold beliefs.

Processing Information Quickly

LLMs are speed monsters when it comes to processing. They can analyze, sort, and synthesize massive amounts of data faster than any human. Need to compare five software tools for your business? Want to understand a dense legal document? LLMs can break it down in seconds.

Similar to the vacation planning example, picture yourself launching a side hustle selling handmade candles. You’re curious about market trends but don’t have time to read 50 blog posts on the subject. You ask an LLM, “Based on the top candle trends for 2025, suggest what styles and themes I must consider for my handmade candles collection. Propose a catalogue, site and newsletter content, and create the script for the first social media posts to spread the word about my new venture.” In less than 30 seconds, it delivers a well-structured, actionable blueprint: the LLM outlines key trends such as eco-friendly production, minimalist design, and innovative scent pairings, and it drafts suggestions for a visually appealing catalogue, a user-friendly website, engaging newsletter topics, and even a series of social media posts complete with catchy taglines. While this output offers a comprehensive starting point, it’s ultimately your insight and personal touch, adjusting the suggestions to resonate with your unique brand identity, that transforms these raw recommendations into a successful, bespoke strategy.

Aiding Automation

When paired with rich context and well-crafted prompts, LLMs supercharge automation. They’re like a tireless specialized executor who follows instructions to a T, handling repetitive tasks so you can focus on the big picture. As long as someone (you) defined well what must be done, what the requirements are, what to avoid, what to include, in which way it must be delivered, and several other aspects that provide context to the task being automated, LLMs will not disappoint. Once again, the powerful duo in action, each one with a different responsibility.

Suppose you run a small bakery and get dozens of emails daily asking about custom cake orders. You set up an LLM-powered chatbot with details about your flavors, sizes, and pricing. Now, it automatically responds to inquiries with tailored options and a booking link. You’re free to focus on baking, not typing. Pro tip: Test your prompts to ensure the bot sounds like you. It is never a good idea to just get what the LLM produce and take it to production. Remember: you should always be in control and ultimately, it is your judgment and personal touches that should determine what the final result is.

The Missing Piece

If we went back to the 1990s and we asked people to imagine a tool so powerful it could answer complex questions instantly, streamlining tasks that once took hours, they would think of a science fiction movie. In the 2012 film Travelling Salesman, four mathematicians dream up a “nondeterministic oracle”, a fictional machine that solves the P vs. NP problem, unlocking answers to impossibly hard questions with ease. While LLMs aren’t that sci-fi oracle (and P vs. NP remains unsolved), they feel like a real-world echo. For automation, LLMs are the missing piece. Feed them rich context (a customer’s query, a dataset, a workflow, a list of specs) and they generate responses or actions that make systems hum. The best part is that anyone can benefit from this new LLM era. A small business owner can automate invoice sorting with an LLM scanning emails, categorizing expenses, and flagging errors, all with a well-crafted prompt. It’s not cracking the code of the universe, but it’s turning tedious work into something fast, smart, and almost magical.

Handling Routine Work

Not every task requires groundbreaking creativity. A lot of “getting things done” involves what I’ll call routine work, that is, standard procedures, quick lookups, or level-1 support. Perhaps LLMs could be described as a super-smart lookup table, delivering accurate, fast results for predictable needs.

Let's see the powerful duo in the classroom. You’re a teacher grading 30 essays. Instead of manually checking each one for basic grammar, you ask an LLM to flag common errors. It won't replace your judgment in any way. It is just handling the tedious work so you can focus on giving meaningful feedback. Once again, different roles working really well together.

Generating Amazing Things

LLMs are called “generative” for a reason. They can produce great things! Text, code, images, even music. They can give wings to your imagination, turning vague ideas into tangible drafts you can refine.

LLMs don't replace inspiration or intention. Let’s say you’re a volunteer organizing a community fundraiser. You need a catchy flyer but aren’t a designer. You tell an LLM, “Generate an image for an upbeat invitation for a charity run. Use a colorful flyer layout.” It spits out a punchy image for the for invite. You take it to Canva so you can add information, tweak here and there and add some personal touches. And just like that, a professional-looking flyer done, zero stress.

The Tools Change, But the Mission Stays

To appreciate what LLMs bring to the table, let’s take a quick step back. Technology evolves and sometimes at a dizzyingly speed. But the essential responsibilities of those who work with it endures. The powerful duo, once again.

In software engineering, decades ago, programmers punched holes in cards to record instructions. Those punch cards were how computers “read” software back then. The programmer’s job was very clear: to solve a problem. That could be something as simple as calculating payroll. Fast-forward to the 1980s, and software lived on floppy disks. By the 2000s, it was CDs, then the cloud. Each shift changed how software was built and shared, but someone was always there, writing the code, deploying it, making it work for others. That someone was the software engineer, the problem-solver.

The tools evolved too. Early coders dealt with assembly language, typing cryptic commands to make computers function. Then came C, C++, Java, Python, each making it easier to tell machines what to do. Now, we’re in the age of AI, where engineers can describe a problem to an LLM and get a working prototype in minutes. Yet, through all these changes, the core remains: someone must define the problem, design a solution, ship it, and keep improving it so users benefit.

This example serves to show us that what we use to solve problems, how we build solutions, and how we deliver them have transformed. Punch cards are museum pieces. But the mission? That’s timeless: understand a need, create something useful, and make it better over time. LLMs don’t replace that responsibility, they amplify it, letting us focus on the big picture while they handle the heavy lifting.

Get Serious About LLMs

Here’s my challenge to you: don’t just play around with LLMs. Get organized and intentional about them. Dive deep. Ask and wonder how they can transform your profession, your industry, and your customers’ lives. What other problems could they solve? What other solutions could they enhance?

Yann LeCun, Meta’s Chief Scientist, once said LLMs aren’t built to be creative or propose novel ideas on their own. That’s our job. We bring the experience, the emotions, the judgments, the dreams. LLMs are the amplifier, taking our sense of wonder and turning it into something real. So experiment. Tinker. Ask big questions. Who can say what’s impossible now?

You can always start small. Try one of the examples above. Then go bigger: imagine how LLMs could reshape your workflow or spark a new venture. The future is no longer some far-off sci-fi dream. It’s here now, and it can be in your hands. Let’s make the most of it.