The reason it's overtaking OpenClaw? It quietly updates its memory of you every few messages, and when it catches you asking for the same task twice, it writes itself a skill for it. So the more you use it, the sharper it gets.

But that only compounds if you set it up right from day one. Give Hermes these three things to start:

• Tell it who you are. Hand it a short profile of your work, your goals, and how you operate. It saves that to its own memory, so every task starts with context instead of a blank slate.

• Set a daily check-in. A morning schedule where it asks you three questions about how you work. Each answer feeds its memory, so it knows a little more about you every day.

• Tell it to be proactive. Ask it to suggest things it could do for you everyday instead of waiting for orders. That's the switch from a tool you operate, to an assistant that anticipates.

I walk through the whole setup in this video.

If you want the exact three-part prompt ready to paste, I've dropped it as a PDF inside our network → here

A member win this week that shows where this leads: Nerrad spun up an AI agent for a Singapore school in 2 days, charged S$500, and it now covers 80 front-desk FAQs a week - not bad! Come join us at the community if you want to start earning from AI too 👉🏻 skool.com/robonuggets

Cheers!
— Jay

That's what /goal does. Claude Code and Codex both shipped it recently, and agents can now run unsupervised for hours on end!

The reason it works? You give the agent a verifiable success condition. Then it loops - run, check, run, check - until the condition is met.

And it works best if your plan is shaped right, so give it these elements:

• Definition of Done. Write one short sentence that says when the work is finished. No fuzzy words like "clean" or "polished". For example: "the page saves a CSV with the current filters."

• Out of Scope. Name at least two things you do NOT want the agent to build. This stops it from wandering off. For example: no mobile version, no login screen.

• A Check Each Turn. Pick one way the agent can test if it's on track - a command, a screenshot, or a small test. Something it can see.

• A Turn Limit. Use 25 turns for small jobs, 50 for medium, and 100 for big ones. Do not go higher.

I show three levels of /goal in action in this video.

If you want to try the plan-for-goal skill yourself - here's a simple prompt you can send to your agentic platform:

I'm giving you a skill called Plan for Goal. Get the files with:

git clone https://github.com/robonuggets/plan-for-goal

Follow the README to set it up.
Recommend how it would best apply to our setup in plain language,
and ask me a few questions to clarify my intent before you install it.

— Jay

P.S: The full Agentic AI Masterclass is inside the RoboNuggets community - takes you from zero to mastery working with AI agents. Come learn with us if that's of interest! → https://www.skool.com/robonuggets

Here's a preview →

The reason? Hyperframes V2 just shipped. It's the open-source video tool from the HeyGen team, and it now handles subtitles, floating cards, and motion graphics — all written as code your agent generates for you.

There are three levels to using it:

Level 1 — Website to video. Paste a URL - Hyperframes scrapes the site, pulls the brand colors, downloads the assets, and gives you a draft video.

Level 2 — Storyboard first. Have Claude render an HTML storyboard before any video gets made. You see the scenes, copy, and shot direction as a webpage. Edit the wording there, then render. (Way better alignment, way fewer wasted tokens!)

Level 3 — Full motion graphics. Cut your raw footage automatically, drop in 3D assets from 21st.dev, key out green screens for rotating logos. This is where the heavy quality lives.

I walk through all three levels with real builds in that video above

If you want to try the Hyperframes Helper skill yourself - here's a simple prompt you can send to Claude Code:

I'm giving you a skill kit called Hyperframes Helper. Get the files with:

git clone https://github.com/robonuggets/hyperframes-helper

Follow the README to set it up.
Recommend how it would best apply to our setup in plain language,
and ask me a few questions to clarify my intent before you install it.

— Jay

P.S: Every other Thursday I run a live bootcam session inside the community where we build setups exactly like this one together. Come build with us if that's of interest! → https://www.skool.com/robonuggets

The reason? GPT Image 2 finally renders long text without typos - 99% of the time. And the bigger unlock: these AI-generated designs are now fully editable, layer by layer!

Three moves cover the full design loop:

• Brief to draft in one prompt. Paste the client brief into Claude. It picks three styles from the Fantastic Posters skill, drops in the logo, and generates the posters.

• Fully editable layers. Drop the final into Canva Magic Layers. It separates text and images automatically - swap text, change the font, move an element, no regenerating.

• Bulk variations in one command. Give Claude one reference design and a folder of 10 photos. It returns 10 posters with the same fonts, colors, and logo - ready to ship across a client roster.

I walk through each move end-to-end in this video.

If you want to try the Fantastic Posters skill yourself - here’s a simple prompt you can send to Claude Code to install it!

``
I'm giving you a skill called Fantastic Posters. Get the files with:

git clone https://github.com/robonuggets/fantastic-posters

Follow the README to set it up.
Recommend how it would best apply to our setup in plain language,
and ask me a few questions to clarify my intent before you install it.
``

— Jay

P.S: The full Agentic AI Masterclass is inside the RoboNuggets community - takes you from zero to mastery working with AI agents so you can start earning from AI. Come build with us if that’s of interest! → https://www.skool.com/robonuggets

Learn More

We just launched the NEW Agents as a Service course inside RoboNuggets

👉🏻 It’s a 24-module program that walks you through building an AI agent service from scratch - from picking your niche, to landing your first client, to scaling up to $10K+ MRR.

And right now, we’re running our BEST offer yet - as it lets you join the premium group for practically free.

***

💎 Limited Offer: Learn and Earn Your Money Back

Join within April as one of our first 1450 members — and if you complete the AaaS course by July 1, we’ll refund your 60-day membership fees back to you 💎

In short: You go from zero to your first AI payday, launch your own AI practice, AND get back what you invested - all if you finish the course.

📌 And btw, you still get our usual member benefits: the Agentic AI Masterclass, the RUBRIC Command Centre, 100+ AI automation templates, a growing community of AI practitioners to connect with, and so much more.

Join RoboNuggets and claim your spot (21 remaining)

Claim your Spot

***

I started RoboNuggets to make AI easy to learn AND earn from.

This offer is our way of rewarding those committed to taking action. And if you do stay in the community by the end of it, then you keep mastering the skill that’s now defining the next decade and beyond.

I’ll see you inside 😉

- Jay

The reason? Karpathy wrote the list of what agents keep getting wrong with code. This file turns it into rules.

The repo is forrestchang/andrej-karpathy-skills. One file. And it fixes all the things Karpathy keeps flagging about working with AI agents:

• Think before coding. Surface assumptions and tradeoffs first. Don’t just start typing.

• Simplicity first. Write the minimum code that solves what was asked. No speculative features.

• Surgical changes. Edit only what’s needed. Match the existing style. Don’t touch adjacent code.

• Goal-driven execution. Turn the task into a success check. Loop until it passes. Example: “fix the bug” becomes “write a test that reproduces it, then make it pass.”

I walk through each rule with real examples in this video.

If you want to try the Karpathy Claude md yourself - here’s a simple prompt you can send to Claude Code:

``
I'm giving you a skill called Karpathy Skills. Get the files with:

git clone https://github.com/forrestchang/andrej-karpathy-skills

Follow the README to set it up.
Recommend how it would best apply to our setup in plain language,
and ask me a few questions to clarify my intent before you install it.
``

— Jay

P.S: The RoboNuggets community has a growing library of Agentic AI setups, skills, and weekly lessons from builders doing this every day. Come learn with us if that’s of interest! → https://www.skool.com/robonuggets

Hope everyone’s new years is off to a good start 😁

We just dropped a new lesson in the channel today — and this one gives you 𝗮𝗻 𝗮𝘂𝘁𝗼𝗺𝗮𝘁𝗲𝗱 𝘄𝗮𝘆 𝘁𝗼 𝗰𝗼𝗽𝘆 𝗮𝗻𝘆 𝘃𝗶𝗱𝗲𝗼 in minutes using an AI creative agent system — built in Airtable and n8n.

And the best part is this: You can now find winning ads in any niche, clone their structure, and repurpose them with your product and your story - without writing prompts manually, and without needing to shoot a single frame of video.

🍳 𝗖𝗵𝗲𝗰𝗸 𝗼𝘂𝘁 𝘁𝗵𝗲 𝗥𝟱𝟭 𝗰𝗼𝘂𝗿𝘀𝗲 here!
And you can find the prompts for that lesson here

📌 I highly recommend this one for agencies, creators, and media teams who want to ship high-quality branded content fast without starting from zero. I also show you new tools in this lesson (𝗹𝗶𝗸𝗲 𝗠𝗮𝗴𝗶𝗰𝗕𝗿𝗶𝗲𝗳 𝗮𝗻𝗱 𝘁𝗵𝗲 𝗦𝗘𝗔𝗟𝗖𝗮𝗠 𝗽𝗿𝗼𝗺𝗽𝘁𝗶𝗻𝗴 𝗳𝗿𝗮𝗺𝗲𝘄𝗼𝗿𝗸) to help you create better and faster!

Cheers,
Jay

1

With the new year upon us, I thought to begin with a build meant to get you started very easily if you haven’t dabbled into offering AI creative services before.

So in this walkthrough, you’ll learn how to build a full AI-powered creative service — 𝘁𝗮𝗿𝗴𝗲𝘁𝗶𝗻𝗴 𝗼𝗻𝗲 𝗼𝗳 𝘁𝗵𝗲 𝗹𝗮𝗿𝗴𝗲𝘀𝘁 𝗶𝗻𝗱𝘂𝘀𝘁𝗿𝗶𝗲𝘀 𝗶𝗻 𝘁𝗵𝗲 𝘄𝗼𝗿𝗹𝗱. And the best part? You can start with it for free, with tools you already use - and just scale it with automation once it clicks 👍🏻

🍳 𝗖𝗵𝗲𝗰𝗸 𝗼𝘂𝘁 𝘁𝗵𝗲 𝗹𝗲𝘀𝘀𝗼𝗻 𝗵𝗲𝗿𝗲!
You can also find the prompts used in this document

***

If you’re kicking off the new year with a goal to start (or grow) your AI creative agency journey, this might be the perfect launchpad 😉

Here’s to an amazing 2026. Happy New Year everyone!

Cheers,
Jay1

I just joined the selfie video trend (just for fun!). And it seems a lot of people are curious about how to make it.

Here’s a quick rundown, and at the end I also give a peek on how I automated the whole thing. Enjoy!

p.s. original idea and base prompts are from the great @pabloprompt. His dragonball video already has 83M+ views!

***

Step 1: ChatGPT for ideas

Ask ChatGPT (or any LLM you use) with ideas for the characters and actors. Chat with it a bit to replace any character or actor if you wish! Here’s the list I did for the Naruto piece:

Step 2: Create the Images with Nanobanana Pro

If you’re doing it manually, I can recommend Kie AI’s Nanobanana Pro instance since they’re offering it the cheapest (9 cents per image, without needing a monthly subscription).

You can of course use Higgsfield or Freepik or Gemini if you already have subscriptions with those.

The base prompt is as follows. Just ask ChatGPT to give you the prompt for each of the characters and use that in Kie AI with your own photo attached!

Create a selfie-style image of this young woman together with Goku from Dragon Ball Z in a hyper-realistic version, portrayed by Brad Pitt. They are on the filming set of the real-life version of Dragon Ball Z. Brad Pitt has long, yellow, very natural-looking hair. Both are casually posing for the camera in a natural, friendly way. Brad Pitt is wearing Goku’s classic combat outfit. The film set includes Dragon Ball Z details. She is not holding anything in her hand. No mobile phones. No phones in hands. She is wearing the same attire as in the reference photo

Step 3: Create the Videos with Kling 2.5

Use Kling 2.5’s first and last frame to generate the clips (Kie AI has the cheapest once again).

Kling is great because it’s very precise - in that when it generates the clips and you stitch them together, the transitions between scenes is seamless. (Veo 3.1 has a problem with this for some reason)

The transition prompt to use is

Selfie-style video. The young woman politely thanks the actor and walks to another area of the set and poses with another actor. She is not holding anything in her hand. No mobile phones. No phones in hands.

When that’s all complete, just stitch your videos together using any video editing tool like Premiere or Capcut and you’re done!

***

That’s the way to do it if you want to do it manually - which will probably take you an hour or two depending on how many scenes you do.

In reality though - I did all 17 scenes in sub-30 minutes because I automated the whole thing using n8n and airtable. It’s a bit difficult to explain in a post, so I have a video here so you can see something like it in action.

And if you want to learn this skill deeper, we teach all of it over at the RoboNuggets Community. It’s mainly for AI-native creative agencies and practitioners who want to create with AI professionally (check that out if that’s for you!)

Hope you learned something new with this one.

And happy holidays!
Jay

We have a new lesson up - and this one’s a gem for those of us working in the fashion niche or other adjacent verticals 🥳

This one shows you how to create 𝗽𝗿𝗲𝗰𝗶𝘀𝗶𝗼𝗻-𝗰𝗼𝗻𝘁𝗿𝗼𝗹 𝗰𝗮𝗺𝗲𝗿𝗮 𝘀𝗵𝗼𝘁𝘀 completely with AI - with cinematic animations between angles of a character using 𝘆𝗼𝘂𝗿 product. And all without needing costly rigs or insane gear setups.

🥚 Check out the tutorial here!

And if you’re after the AI Agent System prompt, you can find that here

We cover:

• How to auto-generate consistent character angles with Nano Banana Pro via Contact Sheets

• Animate transitions between angles with Kling 2.5

• Apply bezier speed curves with a free tool to unlock that buttery smooth motor-rig effect

***

Hope this inspires you to create more.

Cheers,1
Jay 😉

This system is arguably the fastest and cheapest way to create 𝗰𝗶𝗻𝗲𝗺𝗮𝘁𝗶𝗰 𝟯𝗗 𝗣𝗿𝗼𝗱𝘂𝗰𝘁 𝗔𝗻𝗶𝗺𝗮𝘁𝗶𝗼𝗻𝘀 right now - powered by an AI agent system that combines Gemini 3, Nanobanana Pro, Veo3.1, n8n, and Airtable 🥳

🍳 Check out the lesson here!
You can also find the system prompt for the AI Agent here

This one’s PERFECT if you’re into product videography, need a quick concept video for a client, or just building up a demo reel to show off your skills.

It’s crazy how it used to take knowledge of Keyshot or Cinema 4D to make something like this - but now it takes seconds 🔥

***1

Hope this inspires you to create more.

Cheers,
Jay

So here’s an example of a Gentle Monster ad I made…

• created in less than a day

• featuring a real product and character

• with everything you see and hear created by AI
  

🍳 I also show the full step-by-step of how I made this advert here

(and if you need the custom GPT I made, that’s freely accessible by anyone here !)

***

There I show how you can use the latest state-of-the-art models : Nanobanana Pro, Veo3.1, Suno, and ElevenLabs - to deliver quality that can redefine what you can create with AI-generated content.

This is huge if you’re in client work or building a portfolio - you’ll be able to generate polished concepts in less than a day, build a demo reel for your work, and show your skill with studio-grade ads at a fraction of the cost.

Hope this inspires you to start creating too 😉

Cheers,
Jay from RoboNuggets

Share

Happy Thanksgiving to everyone celebrating! I hope you got some time away from screens a bit and got to be with family 😊

2025 has been a wild year for AI, and we’ve had a LOT to appreciate together. And to keep with that theme, here’s a Thanksgiving-styled calendar for 2026 (generated with Nanobanana Pro of course 🍌)

*I gave the prompt at the bottom to make your own with the characters you want, and how you can get yours at 4k!

***

Also wanted to remind you that our Black Friday perk for joining the community is still running this week.

tl;dr - this Black Friday we teamed up with 570+ of the top AI & SaaS tools for big savings and free credits - all yours when you sign up for the annual RoboNuggets community membership!

So when you commit to a year’s worth of AI learning, not only do you get solid education on how to create, automate, and earn with AI (complete with training certification for your CV) - but you also get…

• $5000 in Digital Ocean credits

• $5000 in Microsoft Azure credits

• $2000 in Google Cloud credits

• $1000 in Airtable credits

• 6mos free Notion AI

• and literally hundreds more

…adding up to $3M USD in value when you use them all.

👉🏻 The full list is here (you’ll likely see tools there you already use 😉)

📌 To get all this, you can join as an annual member here:

See plans and perks

*btw - we made the annual plan half price (~$600 off standard rate), to make it easier for long-term AI learners to join.

***

Give it a look if it makes sense for you. If not, all good - you can still get free learning on our channel and in this newsletter.

Cheers!
Jay

Bonus: 2026 calendar prompt and how-to

Here’s a great prompt if you haven’t tried Nanobanana Pro yet and want to see what it can do - it generates well-designed 2026 calendars, with all the dates and text correct! ✨

To use it:

• copy the text below and change the [character request] in the first line, send to chatgpt or gemini to get your final prompt

• send that prompt to Gemini when it’s set to “Thinking” mode (or you can also do it via Kie AI if you need 4k no watermark outputs!)

original prompt found at reddit

******

revise this prompt so it features [Your Character or Theme here] instead. Also adjust the visual style so it fits the characters. give to me in text . Don’t change any parts of the text apart from the characters and style.

Create a colorful 2026 Looney Tunes-themed calendar illustration with an overall 3-row by 4-column layout, each grid representing a month, in a nostalgic and bright style. Date requirements (must be completely accurate): • Use the weekday order of Sun–Mon–Tue–Wed–Thu–Fri–Sat. • 2026 is a common year: February has only 28 days. • Each month must start on the correct weekday as per the actual 2026 calendar: • January 2026 starts on Thursday • February 2026 starts on Sunday • March 2026 starts on Sunday • April 2026 starts on Wednesday • May 2026 starts on Friday • June 2026 starts on Monday • July 2026 starts on Wednesday • August 2026 starts on Saturday • September 2026 starts on Tuesday • October 2026 starts on Thursday • November 2026 starts on Sunday • December 2026 starts on Tuesday Visual requirements: • Looney tune cartoon illustration style with bright colors and a nostalgic atmosphere. • Each month features a different Looney Tune Character theme (Bugs Bunny, Donald Duck, Porky Pig, etc.). • The frame, background, and illustrations of each month match the theme style. • Month titles are in English (January–December). • Dates must be neat, clear, and not misaligned. Layout requirements: • 3-row × 4-column matrix, arranged in monthly order from left to right. • Each grid contains: month title + date grid + small Looney Tune themed illustration. • No repeated or missing months.

There’s a new lesson in our library, and this one shows you how to use one of the most overlooked features of Gemini 3: the 𝗮𝗻𝗮𝗹𝘆𝘇𝗲 𝘃𝗶𝗱𝗲𝗼 node! ✨

tl;dr - if you feed it a YouTube video or shorts link, Gemini can...

• Analyze the full content

• Extract insights + summary

• Write content in 𝘆𝗼𝘂𝗿 tone of voice

• And even help create new videos or publish automatically - if you know how to use it in n8n 😉

🍳 𝗖𝗵𝗲𝗰𝗸 𝗼𝘂𝘁 𝘁𝗵𝗲 𝗻𝟮𝟵 𝗹𝗲𝘀𝘀𝗼𝗻 𝗵𝗲𝗿𝗲!

***

Whether you’re building client content, managing niche media brands, or just want a faster way to outline educational videos - this feature can do 90% of that lifting. And because this is Google - it costs pennies per run.

Cheers!
Jay

Gemini 3 dropped a few days ago and 𝗶𝘁 𝗺𝗮𝗸𝗲𝘀 𝗯𝘂𝗶𝗹𝗱𝗶𝗻𝗴 𝗮𝗽𝗽𝘀 𝘂𝗻𝗯𝗲𝗹𝗶𝗲𝘃𝗮𝗯𝗹𝘆 𝘀𝗶𝗺𝗽𝗹𝗲 for anyone. And the best part: you can try it now for free!

So in this lesson: we show you 𝗵𝗼𝘄 𝘁𝗼 𝗹𝗮𝘂𝗻𝗰𝗵 𝘆𝗼𝘂𝗿 𝗳𝗶𝗿𝘀𝘁 𝗮𝗽𝗽 𝘁𝗼𝗱𝗮𝘆 - from zero to a fully functional app that anyone, anywhere can access. Here we’ll bring together your n8n automation knowledge with Gemini 3 to create great experiences for your clients and users 👍🏻

🍳 Check out the lesson here!

A lot shifted in the AI and creative space this week, but remember that the most important fundamentals stay the same.

Always focus on users and customers, solve their problems, and keep building. I hope the frameworks we’ve shared (including the new VIBE framework shared here!) will help you in that journey 😉

***

Cheers,
Jay

Hey friends!

Thought to let you know that this Black Friday we partnered with 570+ of the top AI & SaaS tools to give our members massive savings and free credits - all available when you become an annual member of the RoboNuggets community by Nov28!

So when you invest in a year’s worth of AI learning, not only do you get world-class education on how to create, automate, and earn with AI (complete with training certification for CV purposes) - but you also get…

• $5000 in Digital Ocean credits

• $5000 in Microsoft Azure credits

• $2000 in Google Cloud credits

• $1000 in Airtable credits

• 6mos free Notion AI

• and literally hundreds more

…adding up to $3M USD in value if you get all of them!

👉🏻 Have a look at the full list here, there’s probably some tools you already use.

You also get access to every new deal we’ll add over the year so that’s pretty cool :)

📌 To get the above and more, you can join as an annual member below

Check the plans and other perks

*btw - we discounted the annual plan to half price (~$600 off standard rate), to give anyone committed to learn AI long-term an even better reason to join

***

Check that out if that’s for you. If not, no worries - you can always get free learning on our channel and through our newsletter (the one you’re reading now as part of the Robo network)

Cheers!
Jay

P.S. if you can’t commit to annual now, you can always try with monthly - and when you upgrade to annual anytime, you still get the discount and perks above 👍🏻

You also get these other benefits that all our members enjoy:

• 🎁 up to $30 USD/month reimbursed for in-person meetups

• 🎁 50% lifetime commissions for every friend you refer to the group

• 🎁 Official ROBO Group™ Certification when completing the Zero-to-Robo course

• 🎁 And of course, access to all our AI Automation templates, a community of 1500+ AI practitioners, and prized learning events (we have a cool one this December!)

See you inside 😉

Hey friends, hope your week’s going great!

Here’s a useful workflow if you need music for your projects 👍🏻

What it does is simple: you give it a music file and a short prompt - and it calls Suno V5 (one of the best AI music models right now), to remix that song to whatever style you want.

So here’s the original RoboNuggets theme:

And here’s the remixed output where I wanted it to be a bit more pop rock:

Sounds pretty good! And it helps a lot if you’re looking to build brand familiarity through music - or if you’re working on films, games, or any media that need variations of a main theme 😉

🥚 Get the template from gdrive here

To use it, you just need to...

• link your credentials to Kie AI (quick tutorial here if you’re new)

• link your file storage in the Output section (in the template we used Box.com)

***

What’s great is the workflow runs at $0.03 cents per song (the cheapest instance of Suno we’ve found, and is super cheap vs having to subscribe to a stock audio site)

And since it’s in n8n, you can plug these nodes into bigger automations if you need to.

Cheers!
Jay from RoboNuggets

FYI: Before you try remixing licensed music, note that Suno actually detects these and rejects them. You can still test it if you want though!

*more info on other music settings you can tweak are on Kie AI’s docs here

Hey folks!

Hope you’re all having a great start to your week 😊

Just thought to share this simple Sora2 watermark remover build in case anyone needs it

📌 Get the template from gdrive here

To use it, you just need to...

• link the gsheet node to this template (link)

• link your credentials to Kie AI (quick tutorial here if you’re new)

***

Kie AI has done a great job with this one. From what I know, they built a custom system that detects and removes Sora watermarks, and it only costs five cents per run so very reasonable.

*Note that I still advocate for labelling your creations as AI-made whenever you get the choice. But this helps if you’re using Sora for commercial work and need a clean, watermark-free result!

Cheers,
Jay

Subscribe now