Uploaded December 2025 | Updated September 2026, 3 weeks ago
Subscribe to the channel for more AI news you can actually use!
For voice dictation that actually works on the devices you actually use, check out Wispr Flow! 👉 ref.wisprflow.ai/aiadvantage
In this video, Igor breaks down Google's new Gemini 3 Deep Think model, which costs $250 to even access (yikes). Watch his breakdown to decide whether Gemini 3 Deep Think is worth it for you or not. Plus he tests the new Adobe apps in ChatGPT including Photoshop and Express, shows off a new way to use Nano Banana, and shares a new way to prompt courtesy of Andrej Karpathy. Enjoy!
Links:
Learn More:
🎓 How-To Save Time With AI: joinaiaclub.com/igor
📸 Instagram: instagram.com/igorpogany
blog.adobe.com/en/publish/2025/12/10/edit-photoshop-chatgpt?sdid=GHMVY429&mv=social&mv2=paid-owned&linkId=100000396641123
chatgpt.com
https://blog.google/products/gemini/gemini-3-deep-think/
https://x.com/GoogleDeepMind/status/1996658401233842624
https://x.com/GoogleDeepMind/status/1996658403167687126/photo/1
gemini.google.com/app
https://x.com/GoogleAI/status/1998442161931534654?s=20
https://blog.google/technology/google-labs/create-presentations-with-nano-banana-pro-in-mixboard-and-more/
mixboard.google.com/projects
https://x.com/karpathy/status/1997731268969304070
https://x.com/xai/status/1997875227884503322?s=20
cdn.openai.com/pdf/7ef17d82-96bf-4dd1-9df2-228f7f377a29/the-state-of-enterprise-ai_2025-report.pdf
anthropic.com/research/anthropic-interviewer
Prompts:
”Create an interactive accurate 3D scene from this sketch of a curved pavilion on top of a museum roof. The structure is an organic, flowing grid of steel with the shown intricate floral geometric cutouts. You must simulate Caustics: I need to see the bright, focused web of light patterns projected onto the concrete floor below, corresponding to the curve of the pavilion lattice above. The shadows should be soft, but the light patterns sharp and intense. Add orbit controls for the time of day, season, and weather so the user can orbit around to see how light travels affects the shadows on the ground. Use a bright, hyperrealistic theme that is visually striking. Give this to me in an HTML file.”
”Create a polished, retro-futuristic 3D spaceship web game contained entirely within a single HTML file using Three.js. The game should feature a "Synthwave/Retrowave" aesthetic. Visual style is a dark, immersive, 3D environment. Gameplay mechanics”
”What would be a good group of people to explore the best way to talk to strangers? What would they say?”
Chapters:
0:00 What’s New?
0:27 GPT vs. Gemini vs. Claude
4:46 Prompting Trick
8:03 Wispr Flow
10:24 Adobe x ChatGPT
13:56 Google Mixboard
16:41 Grok AI Ads
18:25 OpenAI Report
19:55 Anthropic Report
This video is sponsored by Wispr Flow.
Subscribe to the channel for more AI news you can actually use!
For voice dictation that actually works on the devices you actually use, check out Wispr Flow! 👉 ref.wisprflow.ai/aiadvantage
In this video, Igor breaks down Google's new Gemini 3 Deep Think model, which costs $250 to even access (yikes). Watch his breakdown to decide whether Gemini 3 Deep Think is worth it for you or not. Plus he tests the new Adobe apps in ChatGPT including Photoshop and Express, shows off a new way to use Nano Banana, and shares a new way to prompt courtesy of Andrej Karpathy. Enjoy!
Links:
Learn More:
🎓 How-To Save Time With AI: joinaiaclub.com/igor
📸 Instagram: instagram.com/igorpogany
blog.adobe.com/en/publish/2025/12/10/edit-photoshop-chatgpt?sdid=GHMVY429&mv=social&mv2=paid-owned&linkId=100000396641123
chatgpt.com
https://blog.google/products/gemini/gemini-3-deep-think/
https://x.com/GoogleDeepMind/status/1996658401233842624
https://x.com/GoogleDeepMind/status/1996658403167687126/photo/1
gemini.google.com/app
https://x.com/GoogleAI/status/1998442161931534654?s=20
https://blog.google/technology/google-labs/create-presentations-with-nano-banana-pro-in-mixboard-and-more/
mixboard.google.com/projects
https://x.com/karpathy/status/1997731268969304070
https://x.com/xai/status/1997875227884503322?s=20
cdn.openai.com/pdf/7ef17d82-96bf-4dd1-9df2-228f7f377a29/the-state-of-enterprise-ai_2025-report.pdf
anthropic.com/research/anthropic-interviewer
Prompts:
”Create an interactive accurate 3D scene from this sketch of a curved pavilion on top of a museum roof. The structure is an organic, flowing grid of steel with the shown intricate floral geometric cutouts. You must simulate Caustics: I need to see the bright, focused web of light patterns projected onto the concrete floor below, corresponding to the curve of the pavilion lattice above. The shadows should be soft, but the light patterns sharp and intense. Add orbit controls for the time of day, season, and weather so the user can orbit around to see how light travels affects the shadows on the ground. Use a bright, hyperrealistic theme that is visually striking. Give this to me in an HTML file.”
”Create a polished, retro-futuristic 3D spaceship web game contained entirely within a single HTML file using Three.js. The game should feature a "Synthwave/Retrowave" aesthetic. Visual style is a dark, immersive, 3D environment. Gameplay mechanics”
”What would be a good group of people to explore the best way to talk to strangers? What would they say?”
Chapters:
0:00 What’s New?
0:27 GPT vs. Gemini vs. Claude
4:46 Prompting Trick
8:03 Wispr Flow
10:24 Adobe x ChatGPT
13:56 Google Mixboard
16:41 Grok AI Ads
18:25 OpenAI Report
19:55 Anthropic Report
This video is sponsored by Wispr Flow.








![How to Set Up ChatGPT to Work Overnight in 7 Minutes
In this video, Igor shows you how ChatGPTs scheduled tasks work in 2026. While this feature has existed for a while, it has had several updates and new ChatGPT Work integration. If youre wondering what the best ways to use this feature are, youve come to the right place! Enjoy :)
Want to save time, get more leverage, and stop figuring this AI stuff out from scratch? I put the clearest map and support inside the AI Advantage Club: https://joinaiaclub.com/igor
Prompts:
Give me a one-minute briefing on significant generative AI developments from the last 24 hours. Include only real product releases, major model launches, or substantial functional product updates that are already available or demonstrably usable. Exclude rumors, teasers, announcements without a functional product, research-only posts, minor updates, and opinion pieces. Keep it extremely concise. If nothing significant happened, say so plainly.
Check the official OpenAI and Anthropic news and announcement pages for brand-new posts since the previous check. Qualify a post only if it announces a new flagship or named AI model/version (such as a GPT or Claude generation), a major new end-user product, or a significant redesign or relaunch of an existing product. Exclude minor feature updates, pricing changes, API or developer changes, research papers, policy or safety posts, hiring news, and opinion or blog commentary; when uncertain, treat it as non-qualifying and do nothing. Before alerting, check prior automation alerts and sent Gmail messages to ensure this exact release has never been emailed before. For each qualifying, never-before-alerted release, send an email from the connected Gmail account to INSERT EMAIL with the subject `CODE RED — [Company]: [What shipped]`. In the body, write exactly two concise sentences explaining what it is, followed by the direct link to the official announcement. If no qualifying new release exists, do nothing and send no email or notification. How to Set Up ChatGPT to Work Overnight in 7 Minutes](https://i.ytimg.com/vi/gNltRF2XPTg/mqdefault.jpg)

