Do What a Digital Agency Does, But with AI

Plus news from OpenAI, GDM's Genie, Opus 4.1, and more

Author

Martin Musiol
August 06, 2025 • Estimated Reading Time: 9 minutes

In partnership with

This week, we're boosting your web presence.

I'll demo how to use AI to semi-automatically improve your site's crawlability, SEO ranking, and audience discoverability .. and learn important internet insights as a byproduct!

These are the crucial background changes that make a real difference. Make sure you watch the full guide in the video below.

But first, let's cover the big headlines from the past few days in AI tech.

The free newsletter making HR less lonely

The best HR advice comes from people who’ve been in the trenches.

That’s what this newsletter delivers.

I Hate it Here is your insider’s guide to surviving and thriving in HR, from someone who’s been there. It’s not about theory or buzzwords — it’s about practical, real-world advice for navigating everything from tricky managers to messy policies.

Every newsletter is written by Hebba Youssef — a Chief People Officer who’s seen it all and is here to share what actually works (and what doesn’t). We’re talking real talk, real strategies, and real support — all with a side of humor to keep you sane.

Because HR shouldn’t feel like a thankless job. And you shouldn’t feel alone in it.

(✨ If you don’t want ads like these, Premium is the solution. It is like you are buying me a Starbucks Iced Honey Apple Almondmilk Flat White a month.)

Genie 3: The Next Frontier for World Models

DeepMind unveiled Genie 3, a new world model that generates interactive 3D environments directly from text prompts.

It currently hits 24fps at 720p with consistent visuals.

It's easy to see the future here: higher frame rates, better resolution, and on-the-fly rendering in your VR headset based on your input. [more info]

Side remark: DeepMind is cooking in general. Look at these cute storybooks:

OpenAI Releases Open-Weight Models

OpenAI released two high-performance open-weight models, gpt-oss-120b and gpt-oss-20b. Licensed under Apache 2.0, they rival proprietary competitors in reasoning and tool use, are optimized for efficient deployment, and have passed rigorous safety testing. [more info]

And the models can dance with the big guys:

A bar chart titled "Codeforces Competition code" that compares the performance of several AI models based on their Elo rating. The vertical axis is labeled "Elo rating," and the horizontal axis lists the models, noting whether they were tested "with tools" or "without tools." The chart displays the following results, from highest to lowest Elo rating: o4-mini (with tools): 2719 o3 (with tools): 2706 gpt-oss-120b (with tools): 2622 gpt-oss-20b (with tools): 2516 gpt-oss-120b (without tools): 2463 gpt-oss-20b (without tools): 2230 o3-mini (without tools): 2073 The chart visually demonstrates that models generally achieve a significantly higher Elo rating when using tools. The model o4-mini with tools has the highest score in this comparison.

Google Launches Kaggle Game Arena

Google launched the Kaggle Game Arena, an open-source platform for benchmarking AIs through direct competition. It's exciting to see models battle it out in strategic games like chess. You can follow the tournaments live.

See the tournament here.

Last time I checked, it was hot competition! I bet Grok 4 wins. 🙂 

A tournament bracket displaying matches between various AI models in a single-elimination competition. The bracket shows the results of the first round and the pairings for the upcoming semi-finals. First Round Results: Match 1: o4 mini defeated DeepSeek-R1 with a score of 4 to 0. Match 2: o3 defeated Kimi K2 Instruct with a score of 4 to 0. Match 3: Gemini 2.5 Pro defeated Claude Opus 4 with a score of 4 to 0. Match 4: Grok 4 defeated Gemini 2.5 Flash with a score of 4 to 0. Upcoming Semi-Finals: The first semi-final match is between o4 mini and o3. The second semi-final match is between Gemini 2.5 Pro and Grok 4. The winners will advance to the "Gold Medal Match," while the losers will compete in the "Bronze Medal Match." All future matches are marked as "Upcoming" and "TBD."

Claude Opus 4.1 Arrived

Anthropic dropped Claude Opus 4.1. It seems like an incremental update but brings a boost to coding performance to the already best AI coding model. It is particularly good with multi-file refactoring and precise debugging within large codebases.

detailed performance chart for Anthropic's AI model, "Claude Opus 4.1," comparing it against Claude Opus 4, Claude Sonnet 4, OpenAI o3, and Gemini 2.5 Pro on various benchmarks. The chart is split into two main sections. On the left, a text block states that Opus 4.1 advances performance to 74.5% on SWE-bench Verified and improves data analysis and agentic search. Below this, a bar chart titled "Software engineering" visually confirms this, showing accuracy on SWE-bench Verified: Sonnet 3.7 at 62.3%, Opus 4 at 72.5%, and Opus 4.1 reaching 74.5%. The main section on the right is a table with benchmark results: Agentic coding (SWE-bench Verified): Opus 4.1 leads with 74.5%, followed by Sonnet 4 at 72.7%, Opus 4 at 72.5%, o3 at 69.1%, and Gemini 2.5 Pro at 67.2%. Agentic terminal coding (Terminal-Bench): Opus 4.1 scores 43.3%, outperforming Opus 4 (39.2%) and others. Graduate-level reasoning (GPQA Diamond): Gemini 2.5 Pro leads at 86.4%, followed by o3 at 83.3% and Opus 4.1 at 80.9%. Agentic tool use (TAU-bench, Retail): Opus 4.1 is highest at 82.4%. Agentic tool use (TAU-bench, Airline): Sonnet 4 is highest at 60.0%, with Opus 4.1 at 56.0%. Multilingual Q&A (MMLU): Opus 4.1 leads with 89.5%. Visual reasoning (MMMU validation): o3 leads with 82.9%, with Opus 4.1 at 77.1%. High school math competition (AIME 2025): o3 leads with 88.9%, with Opus 4.1 at 78.0%. Footnotes at the bottom describe the methodology used for testing.

So far it did not disappoint. See how well it works in my video below!

Demo: Improving Web Page Crawlability, SEO, and Ranking with AI

We use Cursor + Opus 4.1 to do what you would hire a Digital Growth Agency for.

To see more of demos like this, (I am also open to suggestions), consider subscribing to Premium. 🙂 

I hope you enjoyed it.

Happy weekend!
🙇Martin

Our webpage