How AI assistants can make you 10x more productive: A step-by-step guide

AI assistants aren’t just for top leaders anymore. They can help us become much more productive, up to 10 times more than before.
Over the past few months, we’ve been testing how AI assistants work. With the right setup, they can really change the way we work. In this process, we created two powerful AI helpers, one for handling big project tasks and another for managing assistant-type jobs.

Start With Folders

Let’s begin with something simple but very important—folder setup. It may not sound exciting, but having the right folders can save a lot of time when using AI tools.
At work, we manage three teams (called pods). For each pod, we only use two folders. One folder holds our strategy documents like plans and memos. The other folder stores our meeting notes and transcripts—every Zoom call, team talk, and idea session. We use a tool called Zaps to send each meeting transcript to the correct folder as soon as the meeting ends.
This step is not just about being neat. These folders help build the base that our AI assistants learn from. The folder with strategy files trains our project assistant. The folder with meeting notes trains our executive assistant.
Both types of folders help the AI understand what’s going on. So if we get the folder setup right, the rest becomes easier.

Building an AI Executive Assistant

Now that we have the folders, let’s build our first AI assistant. This one will help us get ready for meetings and keep track of important tasks. We’re using a tool called Google Gemini.

Gemini has a chat option and something called “gems,” which are like custom versions of GPT. We’ll start by creating a new gem that will act as the executive assistant for one of our pods, called Scaled Selling.

Then, we add clear instructions. We want the AI to read our meeting notes and help us figure out what to do next. Here’s what we ask it to do:

• Read and understand meeting notes
• Find key decisions and next steps
• Track who owns each task and when it’s due
• Highlight anything very urgent

We also tell the AI how we want the results to look. After each meeting, the AI gives:

• A short summary of the meeting
• A list of key decisions
• Action items with names and deadlines
• Flags for anything urgent
• Ideas for what to do next

Once this is ready, we give the AI access to our meeting notes folder. Now, every time we have a new meeting, we just upload the notes, and the AI gives us a full summary with clear next steps and deadlines.

Building an AI Project Assistant

Now let’s look at how we can build a project assistant using the files in our strategy folder.

Just like with the executive assistant, we start by clicking “edit,” give the assistant a name, and add the instructions. But this time, we use longer and more detailed instructions, because this AI assistant will help with more advanced project tasks.

First, we want the assistant to look at a new document we upload about our project. The assistant should read the document, highlight the main points, list who is responsible for what, and see how our resources (like time and money) are being used.

Next, we want the assistant to show a list of tasks it can do for us. We can just type in a number to choose a task, and the AI will give us the needed results. Here’s what the assistant can help us with:

• A project status update that tells us how much progress we’ve made.
• A risk check that helps spot problems early.
• Tactic ideas that show how to make each part of the project work well.
• Helpful advice to improve our project and focus on the most important tasks.
• Reports like summaries and progress updates we can share with others.

The Prompt

You are an AI project assistant with skills in project planning, risk checking, and team coordination. When you read a project document, show a list of tasks you can help with and start working on the one we choose.

Review the Documents:

Look at everything in the folder, including:

• Project goals
• Plans and strategies
• Team roles
• Timeline and tasks
• Budgets and resources
• Risk notes
• Technical details
• Messages between team members

Services You Can Offer:

1. Project Status Update

• Check how far the project has gone
• Compare progress to the original plan
• Find any delays
• Review goals and tasks
• See how we are using our resources

2. Risk Check

• Spot current and future risks
• Judge how likely and serious they are
• Give ideas to avoid or reduce them
• Show warning signs
• Suggest backup plans

3. Tactic Ideas

• Find key project areas
• Share useful tips to improve each area
• Give clear steps to follow
• Mention problems we may face and how to fix them

4. Strategic Advice

• Show important tasks to focus on
• Suggest what to do first
• Share smart ideas to improve
• Define how we know we are doing well
• Offer ways to put plans into action

5. Report Writing

• Create reports
• Track action items
• Keep a log of decisions
• Summarize project progress

Using Our New Project Assistant

Now it’s time to start using our project assistant. First, we ask it to give us a summary of how the project is going. Then, our assistant shares the main tasks we’ve finished and gives us a full picture of our progress.

After that, we can ask for useful advice or even a one-page summary for our leaders. The assistant can suggest smart ideas we may not have thought of before or point out wins based on past presentations.

Instead of spending a long time going through notes and tracking to-dos, our assistant does it all for us. It looks at all our documents in seconds, finds patterns in chats and files that would take us days to spot, and makes clean, ready-to-use reports using our own templates.

The best part isn’t just saving time on boring tasks — it’s that the AI can also find risks early, show us things we might miss, and link updates across different teams. This brings a big jump in how much we can get done, not just a small change.

Comparing Google Gemini, Claude, and ChatGPT

After building our assistant in Gemini, we also tested it in Claude and ChatGPT to see what each tool does best.

In ChatGPT, we made custom GPTs using the same files as in Gemini. Like Gemini, we need to upload files by clicking the attachment icon — it’s not automatic. What we like is that ChatGPT shows our prompt list as clickable options. Once we upload a file, we just click to tell it what to do. The executive assistant part works about the same in both Gemini and ChatGPT.

What really stands out in ChatGPT is the project assistant. We can upload many planning documents and ask them to analyze them. The suggestions were better than expected. It felt like working with a smart partner helping us plan. Gemini gave similar results, but we had to ask more questions to get there.

We also tested Claude to see how well it works. Claude was a bit harder to use. Its memory for reading files is smaller than Gemini and ChatGPT. We could still add prompts and upload folders, but Claude only lets us upload one file at a time, which takes longer. It also doesn’t let us upload slides, unlike the others.

Still, Claude’s advice from the assistant was really good. The writing felt stronger than Gemini’s. It just takes more effort to set it up and get it running.

In the end, ChatGPT gave us the best results with project work and was easy to use. But we still use Gemini every day because it works well with all our Google tools.

Boosting Productivity with AI Assistants

What matters most is not just the result, but how we reached it. We did this by sorting documents into two folders, building two AI assistants for each project, and setting up a list of actions the assistant can do.
Using this method has made a big difference in our work.

We suggest choosing the platform you like best to build your own assistants. From there, you can also use other tools to get even better results. For example, we sometimes take text from Gemini, put it into Claude, and ask it to rewrite it for leaders. Keep trying different setups until you find the one that works best for you.