Building Your AI Team
Why treating all AI tools the same is killing your productivity"
Hello friends!
Have you ever had one of those "wait, I've been doing this all wrong" moments? I was talking to a friend about AI, and they were telling me how they love using ChatGPT. I was telling them about all the cool things I was doing and seeing now, not only with ChatGPT, but also with Claude and Gemini. I certainly haven't been shy about using different tools, but what I realized as she wanted to learn more is that I hadn't really formalized how I was using each tool. And that's when I realized I could be more efficient with my time and actually give my friend and you a much better way to work with all the tools. So, here I am, sharing my recipe for how I've been using AI for the past couple of months, and how you can benefit too.
Don't worry I'm not going to go through all the tools. I am going to keep it simple and focus on ChatGPT, Gemini, and Claude. I hope by the end to offer you a new perspective on these tools, rather than always relying on ChatGPT. I should also mention that I am a paying customer of all three. Still, I only started using Gemini seriously, and that's when I realized how each of these tools is truly different and requires a different approach.
If you, like me, treated AI tools as if they were all the same person while using the same prompts, I hope today's email helps you.
Meet Your AI Team
I know AI is not human, but it helps to think of it more like a different colleague on your team – each with its own expertise, personality, and quirks. After countless hours with each system, I now clearly have an archetype to work with in my mind. Here's how you too can start thinking about the big three:
ChatGPT is like that versatile colleague who can jump into any conversation and immediately start contributing ideas. I see them as that solid generalist in the office. Great for brainstorming, getting unstuck, and handling day-to-day stuff. It's accessible, reliable, and honestly feels more flexible when you're working on longer projects. It's no wonder that it's many people's default, and they don't go beyond. It will be fine in most cases, and if you don't know better, you'd never know what you're missing.
Last year, I started using Claude and found that the writing was much more human and creative. Still, I wasn't using it as much until this year, when I began to see them as a very thoughtful project manager type. It's really good at thinking through complex problems step-by-step and writing professional documents that don't sound like they were written by robots. But here's the thing – it hits those context limits faster than you'd expect, which can be frustrating when you're in flow. However, here is a pro tip: use projects and the knowledge you store, as it doesn't count against the context window. I also ensure I have session handoffs for the following context window. This has been a game-changer for me, and it's worth every penny.
Finally, there is Gemini. I was a bit skeptical of it. Conversing with them was very direct. They stick to the plan and are very logical. Since really using it for the last 90 days, I have come to see them as having a brilliant colleague who can see the entire project in their mind with a great memory. More specifically, they are by far the best at development. It absolutely excels at processing massive amounts of information and rarely makes you restart conversations. If you need someone to analyze a huge dataset or understand how dozens of different pieces work together, Gemini's your person - and then hand it off to Clause to build a project plan.
Trust me, these three in concert have changed the way I DO EVERYTHING.
However, one thing I think you need to know, and what no one outside AI nerds really talks about, is the memory problem!
All AI systems have "memory systems", so are smaller or larger than others. Of course, even humans can have "amnesia" on details. Still, if you ever have a lengthy chat with an AI, it appears differently. Sometimes it's hallucinations - like making things up, or not reading all the way through your documents (I used secret phrases - more on that later). You may have heard about tokens or context windows. This is where a chat can only handle so much in one session.
Depending on how you structure your chat, the conversation is only referencable within the chat; that's why I tend to use projects for more lengthy discussions, so I can add a vast library that it can refer to. And one way I check it is by looking at that library, which is deep (usually in the middle) of every document I leave a word in. The AI then has to respond with a specific phrase when we begin, so I know it read everything. AI tends to read the beginning the most, and usually in the end, but the middle is where it gets lazy (kind of like a human).
Suppose you have a large and complex project. In that case, my suggestion is to use Gemini, which has the largest memory capacity and rarely requires restarts. You can use it to synthesize ideas and Claude to workshop and refine them. And then if you need a quick prototype, refine with ChatGPT or the best tool for the gig.
Which brings me to my handy chart and when to use who:
Your AI Team Quick Reference
ChatGPT - Your go-to for everyday stuff
Quick brainstorming sessions
Social media posts
Getting unstuck on ideas
Rapid Prototype
ChatGPT Plus: Higher daily limits (but not unlimited), priority access, file uploads
Claude - Your Project Planner
Writing reports and documents
Project planning
Complex problem-solving
Professional communications
Claude Pro: 5x more usage than free, priority bandwidth, early access to new features
Gemini - The Developer/Analyst
Analyzing spreadsheets or large files
Reviewing code or technical docs
Deep research projects
When you need to process lots of information
Gemini Advanced: 2M token context, integration with Google Workspace, priority support
So, you're likely wondering, Adrian, are you really saying you're paying for three tools? The answer is yes, but they have also become incredibly helpful in my personal and professional life.
For example, my mother-in-law just left for a trip to France and Latvia. She was nervous as she doesn't travel a lot and didn't want to face huge bills using her phone with data back to Canada. I answered all her questions, but then I realized a better way.
I had Claude write a one-page document in her native tongue (Russian) on how and when to use cellular data the moment she left the house, along with a quick-start guide. I gave Claude her carrier information, her phone brand, and they she had wifi in France, a SIM card in Latvia. I gave Claude her itinerary. It spit out a fantastic guide in Russian.
Then I had Claude create one in English for me and my wife to be able to follow along (and check) that everything was accurate. I then had Claude write out a project plan and a document, and I created a custom GPT as a travel adviser that she could use in ChatGPT to converse with in Russian. Time zones are hard between here and Latvia, and we wanted to ensure she had something when she needed it. Piotr is her guide, but he always reminds her to double-check his information; however, he can provide his own advice. She was over the moon, and I was happy to know that we could put her at ease for her trip and give her the tools to get the information she needed when she was unsure.
I should add that I was able to accomplish most of this using free tools, and free tiers are great for testing. However, if you start to really utilize them, you'll hit daily limits pretty quickly. Bottom line: If you're using AI for work, the paid upgrades provide significantly higher limits; however, heavy users may still encounter caps. Choose the tools that your budget allows. Personally, I would opt for ChatGPT Plus, while the free tiers of Claude and Gemini might already up your game.
No matter what you use, I ask you to expand your tool usage, trust me. I see so many smart people – community managers, marketers, and event folks – getting stuck because they're trying to force ChatGPT to do everything. It's like using a Swiss Army knife when you really need specialized tools.
I assure you, once you start working this way, you will have a massive advantage. While everyone else is arguing about which AI is the best, you can be orchestrating teams of AI tools like a conductor making beautiful music.
Ultimately, it's not about finding the perfect tool – it's about recognizing that each tool has its own unique strengths, limitations, and personality, just like building any good team.
So here's my challenge for you: next time you have a complex project, don't default to your usual AI tool. Think about what kind of help you actually need, then pick the right "colleague" for the job.
Your work will thank you for it.
Happy building!
Adrian
P.S. – If you're wondering how to implement this? Start small. Pick one task you usually do with your go-to AI and try it with a different tool. Pay attention to what feels different, what works better, and what doesn't. That's how you start building your own AI team playbook.