🗞️ RUFUS can do what now?💥🚀

Amazon’s AI assistant, RUFUS, just leveled up — and it's not subtle.

🧠 RUFUS can now answer questions about your past orders.
“Where’s that sunscreen I liked?” Boom — RUFUS pulls it up.

It even gives updates like: “When are my dog treats arriving?”
RUFUS: “By 10 PM tonight” (Now chill)

But it doesn’t stop there.🤔 Ask it stuff that has nothing to do with shopping, like: “What do I need to make a smoothie?”

RUFUS goes full recipe-blogger on you (minus the life story) — but the end goal? Getting you to a product you can buy on Amazon.

And for vibes-based queries like: “What do I need for a summer party?” RUFUS brings the whole checklist: string lights, Bluetooth speakers, waterproof snacks — the works.

Oh, and yes — Sponsored Ads have entered the chat. First spotted by Andrew Bell, RUFUS is now surfacing sponsored listings in its answers. Translation? If it looks like a helpful answer, but it's actually an ad... welcome to the new normal.

The shelf isn’t just a predictable, static shelf anymore — it’s dynamic. And RUFUS is curating it.

If you don’t see these yet in your account, just note that it’s being rolled out. The most important thing that brands can do right now to get indexed for RUFUS is to produce large amounts of high quality, intentional content and stop relying on keyword hacks alone.

PPC Ninja has built a bunch of RUFUS specific GPTs to helps with AI-ready listings on Amazon. Help is just an email away. Reach out today: [email protected]

Today’s Cool Tool is Magi-1 by Sand.ai— The World’s First “Autoregressive” Video Model

Most AI video tools still use the same trick for auto generation and it is called “diffusion”.

Start with static, guess all the frames at once, and hope it holds together. It kind of works — until jewelry melts, hands glitch, or your product morphs mid-shot.

Sand.ai is doing something no one else has done (yet):
It’s the world’s first autoregressive video model.

That means it builds each frame in order, like a flipbook — every moment based on the last.

Why does that matter? Because motion stays clean. Objects stay solid.

I tested it with a static photo of a jewelry piece I made a while ago. I want to turn this shot into a video.

Ad Sand.AI did not disappoint. For once, the hand that picks up the earrings looks quite realistic and those earrings? They still look like earrings the whole way through. Pretty cool!

Sand AI isn’t just a new tool — it’s a turning point in how AI video gets made. Less guesswork. More storytelling.

How much you ask? Here are their pricing tiers. You get quite a few credits to try it out for free before you commit.

By the way, I submitted the same prompt to 2 other video generation AI tools for comparison and got some AI bloopers:

OpenAI Sora:

Kling 2.0:

I may be able to salvage parts of the Kling version I think, but the Sora one was quite off.

PPC Ninja is helping brands future proof their listings for AI, helping you build stunning images and videos with AI. Reach out to [email protected] to explore how we can scale your content production across Social media, Amazon ads, Amazon Posts efficiently and affordably.

Agents vs. Workflows — What’s the Real Difference?

Most marketers and operators are familiar with tools like Make.com, n8n or Zapier. You build a flowchart, connect your apps, and automate your work.
Simple. Predictable. Logical.

But now we’re entering the Agentic Era — where automation isn’t just a flow of boxes and arrows, it’s a team of smart AI agents that can think, decide, and act.

Here’s how it breaks down:

Workflows are like Recipes

• Predefined logic

• Great for repetitive tasks

• You control every step
  ✅ Example: “When I get a new Shopify order, send an email and update a Google Sheet.”

Agentic Systems involve Delegation

• Dynamic reasoning

• Agents decide what to do based on context

• You give them the goal, not the steps
  ✅ Example: “Analyze this week’s ads and tell me what to improve” — and the agent fetches data, finds trends, and gives you a plan.

Unlike traditional workflows, agents don’t need hand-holding. They use context, memory, and tools to solve problems — just like a real teammate.

FAQ: Smart Questions About Smart Agents

Q: Where does the agent get its knowledge?
A: From you. You don’t need to “train” it — just give it clean prompts and real-time context (like data, docs, or tools). It uses a pretrained LLM (like GPT-4) to reason through tasks.

Q: Do I need to give it access to my entire system?
A: Nope. You can scope access — just like in Make.com. Let it read from a single spreadsheet, call a filtered API, or write to a draft folder only. Tight scopes = safer agents.

Q: What if the agent messes up?
A: You can build guardrails:

• Add a “critique” agent to review output

• Require human approval before actions

• Use fallback logic (e.g., send an alert instead of failing silently)

Q: So should I ditch my Make.com workflows?
A: Not at all. In fact, smart builders use agents to control workflows.
Example: “Summarize ad data and if performance drops >20%, trigger a Make.com scenario to notify the team.”
That’s automation with judgment — not just logic.

The Agentic Stack Is Coming Fast

Over the past month, we’ve seen an explosion of new agentic tools — from OpenAI’s A2A (Agent to Agent) architecture, to Microsoft’s AutoGen Studio, to open-source frameworks like CrewAI and LangGraph. Even Google’s been quietly building with ADK (Agent Development Kit) and Project Astra. These aren’t just research projects anymore — they’re fast becoming plug-and-play building blocks for real business automation.

The era of “train your model” is giving way to “build your agent.” Tools are the new prompt. Context is the new command line. And the interface? It’s increasingly autonomous.

If you’re still wondering how to use basic workflows without needing to be an engineer, that’s exactly what I’ll show you in the AI for Automations Masterclass on June 5.

The Future of Prompt Engineering: Sam Altman’s Take

Prompt engineering—the practice of writing clear and effective inputs for AI—has quickly become a valuable skill. But according to Sam Altman, CEO of OpenAI, it may not remain a distinct job for long.

Altman sees prompt engineering as a transitional skill. It’s useful today, especially for those working closely with large language models. But over time, as models improve their ability to understand natural language and handle ambiguity, the need for careful prompting will decrease.

This vision aligns with OpenAI’s broader product direction. Altman recently announced that GPT-5 will unify all of OpenAI’s models—eliminating the need for users to choose between different options like GPT-4o, 4.5, o3-mini, preview etc. Instead, ChatGPT will decide how to respond behind the scenes, based on what the user needs.

This simplification means that interacting with AI will become more natural, and less dependent on specialized knowledge. Prompting won’t go away—it will simply become part of many roles, similar to how spreadsheet skills or search proficiency are expected in most jobs today.

For individuals, that means prompt engineering remains a useful skill—but it’s important to see it as part of a broader shift toward AI fluency. For businesses, the focus should shift from hiring prompt specialists to enabling more people across the organization to use AI effectively.

In short: prompt engineering matters now, but its future lies in integration, not specialization.

We’re counting down to June 5th!

Our AI Automation Workflows Masterclass is almost here—designed for marketers, operators, and founders ready to build real automations.

Reserve Your Spot!

Back in January, our AI for E-Commerce Masterclass sold out fast. Now it’s back—with a sharper focus: AI Automations for Scale.

📅 Live on June 5 at 1 PM PST
This is a 120-minute, hands-on session for marketers, operators, and founders who know automation matters—but haven’t yet built their first real workflow.

We’ll go step-by-step inside Make.com, covering:

• How to go from trigger → action → outcome

• The core logic behind every automation

• Where to spot “easy win” workflows hiding in your business

• And how to apply these skills to Zapier, N8N, or Google AppScript

You’ll even get a behind-the-scenes look at how I turn 200+ conference photos into a clean, summarized Google Sheet using AI.

Who is this for?

💡 Intermediate level of AI — no prior builds required
💰 Early Bird price: $99 (goes up to $129 on May 5th)
⚡️ No fluff. Just real stuff I work on.

👉 Sign up here 

Amazon Peak Season Summit

Where: Seoul, South Korea
When: May 22, 2025

Amazon Is flying me to Korea! I'm beyond excited to share that Amazon is flying me—along with 6 of the top names in PPC—to South Korea for an exclusive content shoot. We'll be collaborating on some high-level Amazon PPC content, and you can expect insider insights, behind-the-scenes footage, and strategies straight from the best in the business. It’s going to be in a never before seen “Korean Reality TV Show” style. Stay tuned!

PPC Mastery Summit

Where: Online
When: May 13~15, 2025
Register here 

We hope you liked this edition of the AI for E-Commerce Newsletter! Hit reply and let us know what you think! Thank you for being a subscriber! Know anyone who might be interested to receive this newsletter? Ask them to subscribe here: www.ppc-ninja.com/subscribe. They will thank you for it 💥💪!!

~Ritu