Hello and Happy Tuesday to the regulars and the 435 new subscribers 👋
Fresh from Monday’s podcast: we’ve distilled the most valuable insights and cut through the noise to bring you the highlights.
Let's jump in...
🔎 Preview: In today's issue
- 🤔 Why giving AI access to tools is a huge unlock
- 📺 New Podcast: On the ground feedback from GPT-5
- 🚀 Nano Banana may disrupt Photoshop
Explain Like I'm 5
AI Tool Calling: From Chatbot to Agent
This week's hot topic isn't just another model update – it's about giving your AI superpowers through tool calling.
Whether you're clicking through n8n's visual workflows or writing code against an API, understanding tool calling transforms your AI from a chatbot into an actual assistant that can DO things.
The Butler vs. The Advisor
Imagine you're super wealthy (nice, right) and you can hire two people:
The Advisor sits in a chair and gives brilliant advice:
- "You should book a flight to Tokyo for that meeting"
- "The quarterly report needs updating with last week's numbers"
- "Someone should order more coffee for the office"
Smart Yes. Helpful Sort of. But they never actually DO anything.
The Butler has the same intelligence but can actually act:
- Books your Tokyo flight while you're in another meeting
- Updates the quarterly report and emails it to the team
- Orders coffee and schedules delivery for tomorrow morning
That's the difference between AI without tools (advisor) and AI with tools (butler).
One talks about what should happen. The other makes it happen.
The Magic Moment: When AI Becomes an Agent
An "agent" is just AI that can take actions on its own. Here's the beautiful part... you don't micromanage it.
You say "Handle my travel to Tokyo next week" and it:
- Checks your calendar for conflicts
- Searches for flights
- Books the best option
- Adds it to your calendar
- Emails you the confirmation
You gave ONE instruction. It took FIVE actions. That's agency.
How It Works (Visual vs Code)
🎨 Visual Tool Calling (n8n, Make, Zapier)
In platforms like n8n, you literally drag and drop tools onto your AI agent – like giving your butler a tray of abilities. Connect a weather API node, and boom – your AI can check weather. Add a Google Sheets node, now it can update spreadsheets. Each connection is a new skill.
💻 API Tool Calling When coding, you define available tools in your API request:
tools: [{ name: "check_weather", description: "Get current weather for a location", parameters: {...} }]
The AI decides WHEN to use tools (just like a butler knows when to go to the kitchen), calls them, gets results, and continues the conversation with real data.
Enter MCP: Your Tool Organisation System
Here's where it gets interesting. As you add more tools, things can get messy fast. Imagine your butler having 50 different abilities but no organised way to manage them. That's where Model Context Protocol (MCP) comes in.
MCP is like a tool cabinet for your AI. Instead of scattered tools everywhere, you have:
- Organized drawers: Group related tools together (all Google tools in one place)
- Easy updates: Change a tool once, every AI using it gets the update
- Simple sharing: Built a cool tool Share the whole "drawer" with your team
- Clean connections: One connection gives access to multiple tools
We recently covered how to build your own internal MCP server in n8n – turning your custom business tools into a professional toolkit your AI agents can use instantly.
Watch our How to build your first MCP tutorial
☕️ Knowledge Shots
1️⃣ The "Loop" that changes everything: AI agents can check → decide → act → verify → adjust. This loop is what makes them feel autonomous. They're not just responding – they're thinking and adapting.
2️⃣ Tools aren't just for data: Your AI can send emails, create calendar events, update databases, generate images, or even control smart home devices. If there's an API for it, your AI can probably use it.
3️⃣ Platform differences:
- OpenAI: Calls them "functions" or "tools"
- Anthropic (Claude): Uses "tools" with MCP support built-in
- Google (Gemini): "Function calling" with similar structure
- n8n/Make: Visual "AI Agent" nodes with tool connections
🧠 Why This Matters Right Now
AI is at a turning point: it’s no longer just suggesting ideas. It can take action. One prompt can now trigger dozens of automated tasks, moving AI from “neat toy” to a practical assistant.
But while tool calling and orchestration unlock real productivity, you still need humans in the loop. AI can handle the busywork, but judgment and oversight remain essential.
That's the revolution happening right now. And you can build it today.
📺 PODCAST: Claude Code, Nano Banana GPT-5 Launch
In this week’s Create With podcast, James and Kieran dig into practical ways non-developers can ship serious features with AI.
We share what finally “clicked” with Claude Code, how to spin up simple Bubble plugins (goodbye $5/mo ICS generators), first impressions of GPT-5 (the good, the weird, and the slow), why object-preserving image edits could upend Photoshop, and how we’re automating the Create With site - events, tools, blog posts end-to-end.
▶️ Watch Video
🔊 On your radar
Keep your finger on the pulse of what we think is hot in the space right now 🔥
⭐️ Nano Banana: Google’s Secret AI Image Editor
👉 Why it matters: Nano Banana is Google’s next-generation image editing AI now available in Gemini. It lets you edit photos with natural language, no Photoshop skills required. Early testers say it’s fast, preserves identity across edits, and could change how creators, marketers, and teams handle visual workflows.
⭐️ #BubbleMobileContest: Thousands of Apps, Real-World Impact
👉 Why it matters: Bubble’s new native mobile editor has inspired thousands of submissions in the #BubbleMobileContest, showcasing what’s possible when no-code meets mobile development. One standout entry lets users appeal parking ticket fines by snapping a photo and submitting an excuse—practical innovation at its best. With judging underway, you can browse the hashtag on X for more clever ideas and fresh mobile app prototypes.
⭐️ Softr databases has integrated with Make.com
👉 Why it matters: On Day 4 of the #10DaysOfSoftr challenge, Mariam Hakobyan showcased how to supercharge Softr Databases with Make automations. Rather than using webhooks or a custom connection you can plug in directly to to a Softr database
🔗 Bonus links for subscribers
🔒 Planmode on Claude Code is not obvious to find. Use the internet to help you
🔒 n8n has new pricing packages: catering for enterprises
🔒 Cool story from Bubble and ThinkCater
🗓️ Events to Take Note Of
🇬🇧 freeCodeCamp | AI Edition - Creating your First Local AI Agents (Aug 27th)
🇨🇦 We'll be in Montreal for a CreateWith Meetup with Benoit (September 11th)
🌐 The Oscars of AI film, music and games just dropped: flex your skills to take home the ultimate prize in November! Introducing The Chroma Awards - expect free tool trials starting from Sept 1st
🖱️ Discover more events on the new Create With events directory!
💭 Final Thoughts
Last week, we hosted our first members-only office hours: a relaxed, private space for sharing both wins and challenges. When it came to wrangling AI, the session felt equal parts practical discussion and group therapy. Thank you everyone who came along!
See you next week!
With good vibes ✨
{% poll 28123 %}{% for option in poll.options %}{% endfor %}
{{ poll.question }}
[
]({{ option.url }})
[{{ option.value }}]({{ option.url }})
{% endpoll %}
{{ address }}