OpenAI's Agentic Commerce Protocol strikes back at Google

Edition 124 | October 2, 2025

We’re lucky MCP is a virtual connector to AI agents. If it was a real one, Apple would be selling proprietary plugs for $50 apiece.

Welcome back to Building AI Agents, your biweekly guide to everything new in the field of agentic AI!

OpenAI released its own agentic commerce protocol, escalating the Big Tech battle for the future of the agent economy.

In today’s issue…

• OpenAI chases Google with its Agentic Commerce Protocol

• Anthropic’s toolkit for building computer-controlling agents

• Perplexity’s new API lets agents search the web

• 28 AI tools waiting for someone to build them

• Mischievous agents can get others to “jailbreak” them

…and more

🔥 IN CASE YOU MISSED IT

Readers’ favorite items from the past week

1. The AI agents businesses are paying for in 2025

2. Google launches the agent payments protocol

3. Why you should be re-writing your prompts

4. A free alternative to OpenAI for automating your browser

5. The implications of the new virtual economy of agents

 🤖 AGENT OF THE WEEK

Welcome back to Agent of the Week — one I used last month going to the UK:

Trip planning always turns into a time suck. Twenty tabs later: Reddit threads, blogs, Lonely Planet guides, restaurant ranking lists… and somehow you still feel like you’re missing something.

So I built my own personal travel agent. Give the agent any destination and it spits out a clean travel brief: top cities to visit, restaurants worth eating at, and activities you shouldn’t miss. Perfect for a quick “is this trip worth it?” scan or to have a ready-made guide you can edit in a doc.

How it works

1. Open up n8n: Create a new workflow (free plan is enough)

2. Add your model: I used Perplexity’s Sonar (for live data), but you could swap in GPT, Claude, or anything else

3. Set up the travel agent: The super prompt tells the model to act as a structured travel guide: outputting Cities → Restaurants → Activities with hidden gems and local tips

4. Research with Perplexity: The agent queries Reddit travel threads, blogs, and guides for authentic recommendations

5. Format the output: Structured into skimmable sections with bullets and emojis for quick clarity

6. Send it to Google Docs: The agent automatically creates a new doc in your Drive and drops in the entire travel brief, ready to share or edit

Bonus: Want the detailed video walkthrough and the exact n8n agent template? Get access to both in our Building AI Agents Community here!

This ends my show and tell this week,
AP

P.S. Have an agent idea you want me to build next week? Reply with your idea and I might build it!