Gemini can turn any book into an interactive mindmaps

You can do it for books of hundreds of pages thanks to the massive context window and with a single prompt.

Perfect for those who memorize better visually (demo and prompt below):
1. Upload your book

Open Gemini and upload your book in PDF format.

You can also use Google AI Studio by selecting the 1.5 Pro model.

Reasoning models are becoming the norm

So a friendly reminder that OpenAI has released a guide to writing prompts for them.

Few important points and example: The most important thing is to use delimiters, for example XML tags that look like this:

<context>
Your text...
</context>

You should also nest tags, for example: <outer><inner></inner></outer>

A concrete example below:

Google has just launched an AI feature capable of planning, reasoning and searching the web.

All this is done in a single prompt using Deep Research.

This is literally the first reasoning model that has access to the internet. Game-changer.

How to use it below 🧵 1. Select model

Open Gemini Advanced and select “Deep Research” from the list at the top.

Then type your prompt. Example:

“Research all the possibilities for running a small AI model (LLM/SLM) locally on a Chromebook, without an internet connection.”

Stop using Make or Zapier

You can use Replit to create and deploy any Python script instead.

- It's faster
- Much more flexible
- Lower cost
- You're in control

Just describe what you want the Agent to do (e.g. scrap a website).

Run and deploy in one click with Replit 🧵 1. Write your prompt

Open Replit Agent and write your prompt.

E.g. if you want to scrap a website:

“Write a Python script to scrap the [Website Name] documentation, starting with this URL: [Your URL]”

I just built my 5th app using Replit AI and Agent.

Not a single line of code written by myself as I don't know how to code.

You have to follow the right methods and you can even create a full SaaS.

Here's how to create anything without coding 🧵 1. Define your Minimum Viable Product (MVP)

- What core features do you want your app to have?
- To solve what problem?
- With which UI?

Keep it simple for the initial prototype.

Wow can't believe this AI model runs so fast locally on my $250 phone.

No internet connection. 100% private.

And the power of the model is surprisingly good for its small size.

iOS and Android tutorial below if interested: 1. Install PocketPal

Search for PocketPal on the App Store or Play Store and download the app.

100% free and open source.

I’ve finally set up ollama to run any open source AI model locally and for free.

It only took me a few minutes, contrary to what I thought.

Glad to have an offline and private chatbot even on my low-powered laptop.

(Windows and macOS tutorial below if interested) 1. Install Ollama

Head over to the ollama website:

Just download the installer for your OS. ollama.com

Gemini's new Gems are WAY better than GPTs

You can create custom AI chats that have access to YouTube, Gmail or Drive.

This unlocks awesome new possibilities. Here's how to create one: 1. Access the Gem Manager

Open Gemini Advanced and click on the new Gem icon at bottom left.

Then click on New Gem.

I built a full Chrome extension without knowing how to code in less than an hour.

It lets me speak in my native language and it writes directly in English thanks to Whisper.

Here's how you can build any app using AI: 1. Choose the right AI chatbot (LLM) for your project

If your app is simple, you can use ChatGPT 4o or Gemini Advanced.

The most powerful LLM to write code today is Claude 3.5 Sonnet: perfect for more advanced projects.

For example, I will create a Chrome extension that allows me to speak in French while writing in English. So I will use Claude.

This is the best way to create custom AI agents.

Mistral has just released Agents that allows you to easily use and customize their models for your tasks.

Here's how to create your own: 1. Access the Mistral platform

Head over to Mistral AI's La Plateforme at mistral .ai.

If you're new, create an account and log in.

I started automating all the tasks on my computer using ChatGPT.

Organizing 80+ files took me only 3 prompts and 5 minutes.

Here's how you can do it (Windows, macOS, Linux…): 1. List the files in a folder

Open ChatGPT and describe that you want to list all the file names in a specific folder using a command:

Prompt:
"How do I copy all the file names in a folder in Windows 11 using a command?"

ChatGPT will provide a suitable command like `Get-ChildItem -Name | Set-Clipboard` for PowerShell.

OpenAI has deployed interactive charts and tables in ChatGPT.

You can now interact with your data and process it with GPT-4o.

Here's how it works and what you can do with it: 1. Upload your data

Make sure you select GPT-4o as the template.

Then upload the file with your data (xlsx, csv, ...).

ChatGPT analyzes the columns - all you have to do is enlarge the table.

This is huge.

Gemini can now analyze up to 1500 pages.

Google has integrated Gemini 1.5 Pro which handles up to 1M tokens.

It allows you do great things you can't do on other AI chatbots 🧵 1. Document Analysis

Analyze large documents for quick summaries, key insights, or comparisons.

Note:

You can upload files from your computer or from your Google Drive.

Microsoft has made Copilot GPTs even better

You can now write instructions up to 8000 characters long (2k before).

This opens up possibilities that didn't exist before 🧵 1. Create a GPT Copilot

Reminder:

- Create a Copilot GPT by clicking on "See all Copilot GPTs" on the right

- Then "Create a new Copilot GPT"

You need a Copilot Pro account to create Copilot GPTs, but anyone can use a GPT even with a free account.

Meanwhile Google has updated Gemini

Everyone can now use YouTube, Maps, etc. extensions.

You can chat with YT videos, plan trips, manage email and more.

This was only available to a few countries before.

Details below: To activate extensions:

- Go to settings
- Click on "Extensions

All you have to do is check the ones you want.

Paradigm shift.

Copilot in Windows now performs actions for you.

Tasks that used to take several minutes can now be done with a prompt.

Rename dozens of PDFs, create a ready-to-use Excel file, and so on.

Some examples and how to enable it below: 1. Enable Power Automate

Open Copilot in Windows and go to Plugins.

In the list, enable Power Automate.

Roll-out is gradual (I personally use the Windows Dev channel).

You can change channel or wait if it's not yet available on your system.

Absolutely stunning.

Llama-3 8B was the most powerful small AI model a few hours ago.

Microsoft has just released Phi-3 Mini which is better in almost every benchmark.

You can use it locally for free as it's open-source (details below): 1. Download LM Studio - it's free, open-source software that lets you run any open-source AI model.

Go to , then download and install the one for your OS.

lmstudio.ai

Microsoft is turning Windows laptops into AI PCs with Copilot.

I've been using the new feature that reads any file for a few days.

It's the most powerful update deployed so far.

3 quick examples to understand why: Analyze uploaded documents

Copilot in Windows can now process and understand various file types.

It allows it to:

- Extract relevant information
- Summarize key points of the document
- Understand the structure and content

Mistral AI has released the most performant open model

Uncensored, fluent in several languages, strong in math and coding, ...

It's free to use by anyone without a login or account.

More details and how to access it below 🧵 1. Mixtral 8x22B features

- Fluent in English, French, Italian, German, and Spanish

- Strong mathematics and coding capabilities

- 64K tokens context window

I've just received access to Meta AI.

Its reasoning, web search and image generation capabilities are surprisingly good.

I wasn't expecting anything, but I'm going to keep using it.

Here are a few examples and how to access it 🧵 1. Writing and summarization

Meta AI can write content and summarize texts just like any other AI chatbot.

But I was surprised to see that it even accepts very long prompts:

Gemini is just amazing for creating charts.

You can search for reliable data and visualize it easily.

Here's how to generate and even modify charts using Gemini (even with the free version) 🧵 1. Search and find data

- Open Gemini (free version or Advanced if you have it):
gemini .google. com

- Enter the prompt below to find and structure data

As Gemini is connected in real time to the Internet, it will give you up-to-date data and links to access it.