Learn to create with AI NoCode.

Hey, happy Tuesday. It's a great day to read the Create With newsletter! ✨

And welcome to the 332 new subscribers since last week 👋

Let's get you up to date with the latest in the AI and visual development space! 💡

🔎 Preview: In today's issue

🪟 Token Windows explained
🎓 Tutorial: MakeAI vs Lindy for creating AI Agents
⚡️ Build Weekend Registration is Open!
📡 On your radar: Airtable and Make add new AI capabilities 🔥

**😣 Explain like I'm 5 (Jargon Buster)

Token Windows

🪟 The token window (or context window) is the short-term memory space of an AI model. Everything you say — and everything the AI replies — gets converted into tiny chunks called tokens (usually parts of words). The AI can only keep a certain number of these tokens in memory at once.

❓ Why it matters
This window affects how well an AI can understand your current task. Once it’s full, older information starts to disappear. That’s why long chats may feel like the AI is "forgetting" — it literally doesn’t have room to remember everything. You’ll often be told: "Please start a new chat" when this memory space hits its limit.

**🧠 Analogy:**
Think of the context window like a super-powered clipboard. You can keep stacking in bits of a conversation, images, transcripts, and files — but only up to a certain limit. After that, older stuff gets pushed out. Hence why it's best practice to start a new chat if you are changing the task slightly

**🤓 Here’s what’s cool:**
This “memory” doesn’t just come from typed text; but it ultimately ends up as tokens.

You can upload files, copy-paste documents, or even input images, and those get converted into tokens too.
Some advanced tools also process audio, code, or even do live internet research — all of which are added to the token window behind the scenes.
Google Gemini has one of the biggest token windows available — reportedly over 1 million tokens — so it can look at huge files or complex multi-part prompts all at once.
ChatGPT-4 (depending on the plan) goes up to 128,000 tokens, which is roughly the equivalent of 300+ pages of content.

🦾 Pro tip
If you're getting near a model's token limit:

Ask it: "Summarise everything so far in 500 tokens."
Copy that summary and paste it into a fresh chat.
Continue from there.

This compresses the context and preserves relevance.

🔊 On your radar

Keep your finger on the pulse on what's hot in the space right now

⭐️ Airtable AI is in Beta

👉 This matters because__: It can generate formulas, automate tasks, and assist with data organisation—all through natural language prompts. This move pushes Airtable closer to becoming a fully AI-powered workspace for teams.

⭐️ Make releases AI Agent (in Beta)

👉 This matters because__: Make are a primary player in the automations space and this keeps it competitive for you as a builder. It's a little buggy (makes sense) and UX could be improved; but the nuts and bolts are there. Very much ready to test out on non-business-critical tasks. The agents can be shared across workflows (neat) and has a global system prompt for maintainability but allows customisation if needed. It's not guaranteed you'll have access if you try

⭐️ Windsurf (AI Assisted Code Writing IDE) gives you access to GPT4.1 and 04-mini for free

👉 This matters because: Getting your hands dirty with code writing is hard. Windsurf was relatively straight forward to wrap our heads around and now you can use frontier models for free, at least until the end of hte week.

Extra bytes

OpenAI release o3 and o4 but it's supposedly hallucinating a lot more
Deepmind thinks we're at the next stage of ai development; through streamed 'experiential' learning, without human pre-judgement. The approach would allow AI to learn with extended real-world interactions and feedback, and improve over time!
With Grok Studio, Grok has feature parity with ChatGPT

️ Build Weekend Registration Is Open!

Registration is open now!

📣 In collaboration with 100 School we're bringing you ⚡️BUILD WEEKEND 2025! Unlock your ideas in just 3 days! Join hundreds of other builders worldwide and win prizes for your ideas! Sign up now at BuildWeekend.com.

🗓️ Friday 16th to Sunday 18th MAY, 2025

This event is open worldwide! No technical knowledge or coding experience required - just bring your ideas and get ready to bring them to life.

That's me! I want to make something!

Tutorials:

This week we're bringing you two AI Agent tutorials: MakeAI vs Lindy. Get an overview of the build process for both with examples of what you can make.

9x cover the first look at Make's AI Agents. Jan covers how to make a Lead Research Agent trigger to get LinkedIn Profile details, a Google Search and extract content from a website.

Watch it here

Lindy also allows you to create AI agents with an intuitive canvas builder. In this video you'll get an overview of the Lindy Builder with a calendar scheduler and knowledge base example.

Watch it here

The big idea

A centaur is a mythical half-human half-horse. Turns out some AI researchers are using this term to describe to how humans work with AI.

The big idea (in plain English)

When chess grand‑masters pair with machines they win more than either alone; the key is a clean hand‑off where the human handles strategy and the machine tactics.

Think of a centaur as “you + an AI sidekick working as one brain.” Let the software crunch data, recall facts, draft first passes; you step in where gut feeling, context and persuasion matter. Studies show this tag‑team often beats the best solo human or the best standalone algorithm.

How to put it to work right now

List your daily tasks → circle the repetitive/rule‑based bits. Hand those to ChatGPT, Copilot, or no‑code automations.
Save your energy for judgment calls: strategy choices, negotiations, creative angles, client empathy.
Build a “handoff ritual”: AI drafts → you skim, tweak, and decide. Aim for <5‑minute handoff friction.
Capture feedback: when your tweak differs from the AI, write a one‑liner why; feed that back as a new prompt or fine‑tune example.
Review monthly: anything the AI now does well with zero edits Automate fully and climb to the next layer of value.

This edition of Create With is brought to you by... Momen

Momen.app

Speaking of AI agents, we wanted to take a moment to thank our sponsor Momen!

Momen is a full stack web app builder allowing you to build complete AI Agent apps with frontend, backend and database.
Payment integration Data permissions API capabilities Templates SEO controls It's all built in.

⚡️ Build Your First App With Momen

They'll be at the Create With Conference to answer all your questions on the 22nd of May.

Final thoughts

Todays issue is dense so we'll leave you with something light.

OpenAI reportedly spends tens of millions dollars processing "please" and "thank you" messages.

We're going to take a guess it's mainly the Brits (and Canadians) spending millions of dollars saying 'sorry' to OpenAI.

With good vibes 👋

✨

[]({{ affiliate_url }})