# This Week in All Things AI - Week 37-2025

> Sunday 7th September 2025 to Saturday 13th September 2025

**Published by:** [This Week in All Things AI](https://paragraph.com/@twiata/)
**Published on:** 2025-09-13
**URL:** https://paragraph.com/@twiata/this-week-in-all-things-ai-week-37-2025

## Content

As we head into the end of Q3 2025 and start into Q4 2025, I suspect that there might be some surge in updates to models from the frontier labs [Google, OpenAI, xAI, Anthropic] and the mainland Chinese giants [Alibaba/Qwen, ZAI, Deepseek] also expected to keep pushing the envelope. I also expect more content surfacing as users experience Replit Agent 3 which you will read more further in this edition Readers are welcome to share this publication link to those in your family, friends and acquaintances who have a curiosity in knowing more about AI tools/services/use cases which are rapidly evolving. For those who use RSS readers to aggregate their reading of news such as the delightful NetNewsWire [MacOS,iOS, iPadOS] or Feedly [web, iOS, Android], this weekly newsletter is also available via the following RSS feed https://paragraph.com/api/blogs/rss/%40twiata Below is my personal website which aggregates links to many of my socials as well as content and community that I curategoolamabbas | Instagram, TikTok | LinktreeChief Knowledge Officer @ Animoca Brands. Also assists in Growth Marketinghttps://linktr.eeContinuing to use a tabular format to show the digest of posts on days which have had multiple posts to isolate them individually. The following messages were posted on the 'All Things AI' Telegram group from Sunday 7th Sep 2025 to Saturday 13th Sep 2025Sunday 7th September 2025Monday 8th September 2025Poster: Matias Hi folks! Any recommendations for tools to create powerful Google Slide pitch decks? I have the content ready in the Google Slide, but it's raw and not graphics. Been trying a few tools, but didn't find anything particular great. Happy to pay for subscription. Reply-by: Yusuf Matias, I thought Gemini was available via Google Slides though that could depend on your Google Workspace or your Google AI subscription plan https://www.perplexity.ai/search/google-slides-gemini-ai-PBYBO.2uTqeK_oadIeUowA Some alternatives I'm aware of though never tried any one of them in sufficient depth to give any worthwhile recommendation. I've a leaning towards Gamma though https://gamma.app/ https://slides.com/ https://www.canva.com/create/ai-presentations/ https://www.slidesai.io/Tuesday 9th September 2025 1New features rolling out in Google's NotebookLM product today -Generate study materials including Flashcards and Quizzes -Create more types of Reports -A new "Learning Guide" feature that acts like a personal tutor, coming next week -Curated notebooks from OpenStax -New types of Audio Overviews6 ways to use NotebookLM to master any subjectThis semester, students can use NotebookLM to instantly generate flashcards, quizzes, professional reports and more.https://blog.google2Cognition makers of the Devin AI software engineer and now owners of the Windsurf agentic IDE raised $400M led by Founders Fund at a $10.2B valuation, up from $4B in March; its coding tool Devin's ARR grew from $1M in Sept. 2024 to $73M by June Two of their early investors , Christian Lawless (Conversion Capital, Founder & General Partner) and Emily Cohen (Neo, Partner), joined them full-time. They also got swyx (Shawn Wang) of Latent Space, AI Engineer to join themCognition | Funding, growth, and the next frontier of AI coding agentsCognition has raised over $400M at a $10.2B post-money valuation led by Founders Fund. Other existing investors are similarly doubling down including Lux Capital and 8VC (who jointly led our previous round), Neo, Elad Gil, Definition Capital, and Swish VC. We're also joined by new investors including Bain Capital Ventures, Hanabi Capital, and D1 Capital.https://cognition.aiCognition: The Devin is in the DetailsMy thesis for the future of software dev agents. This is a hastily written blogpost done on not a lot of sleep, so pardon poor pacing and structure and typos and mistakes but just getting it out there.https://www.swyx.io3Replit teasing a major upgrade coming to their platofrm Livestream scheduled for Wednesday, September 10, 12:00 PM - 1:00 PM US Pacific Time (PDT). Replit ⠕ @Replit We have been working hard on something exciting- designed to change the way you build Sign up for the livestream below to learn more 456 3:40 AM • Sep 9, 2025 Ryan Mulligan @ryantm Agent is about to get a lot more autonomous and easier to use. Replit ⠕ @Replit We have been working hard on something exciting- designed to change the way you build Sign up for the livestream below to learn more 156 3:52 AM • Sep 9, 2025 Scott Kennedy ⠕ @stkenned Huge release. This took months to pull off and it's going to wow you. Replit ⠕ @Replit We have been working hard on something exciting- designed to change the way you build Sign up for the livestream below to learn more 98 4:19 AM • Sep 9, 2025 Wednesday 10th September 2025Poster: Roc Zacharias I was asking these to Yusuf in DMs, but figured it's better to ask here so everyone can learn together :)Hey bro I want to start saving all of my business meeting calls, ingesting all of my Telegram conversations, etc to start building a ILM (Individual Language Model) and/or digital twin. Do you have any recommendations? I am thinking it might not matter what I use for the time being as I just need to record the content for now, and then can ingest it into a personal model later right? If there are good tools to use it with now that's even better, but I want to be careful about someone having my personal conversations as data Actually I'll ask this in your group :)Basically I can only recall probably 1-5% of facts from my calls over the years. I want to save all of my conversations, emails, TG messages, whatsapp, LinkedIn, personal calls, and maybe even use a pendant to record my in person conversations. My memory is getting worse with age, and I want to be able to recall all of my conversations. For example, I could ask my twin or ILM: "I'm going to Japan in 3 days, who do I know in Japan that I may want to meet with during Eth Tokyo?" Reply-By: Mau I got recommended this one by my friend who’s in coaching: https://www.plaud.ai/products/plaud-notepin and he said it works well, is discete, and integrates with all his LLMs (mainly syncing transcripts straight to ChatGPT). The only thing I’m not sure about is how compliant it is for sensitive data sharing. But thinking it’s quite similar risk as using ChatGPT/AI tools already. Only activates when you want to use it Reply-By: Yusuf This seems to be a hard problem because you are asking for a tool which can access different data sources and possible across different modalities. From a pendant PoV, the closest I've seen but never used is this product from Limitless AI https://www.limitless.ai/Thursday 11th September 20251via Replitwe are proud to introduce Agent 3, our most autonomous Agent yet 🔥 What's new with Agent 3: ⚡️Tests & fixes the App it builds automatically using a browser- checking buttons, forms, links and APIs. ⏲Runs for 200+ mins autonomously as it is building, testing and fixing your app- eliminating the need for manual oversight. ⚙Builds Agents and automations to automate complex & repetitive tasks. You can now build Slack or Telegram bots or workflow automations integrated with Notion, Linear, Sharepoint, Dropbox & more in plain English. Agent 3 starts rolling out to all users today Amjad Masad @amasad AI agents can prototype apps… But shipping real software takes hours of testing, debugging, and refactoring. Agent 3 is 10× more autonomous — it keeps going where others get stuck. The “Full Self-Driving” moment of software. 4,725 11:31 PM • Sep 10, 2025 Replit - Introducing Agent 3: Our Most Autonomous Agent YetWe're excited to introduce Agent 3-our most advanced and autonomous Agent yet. Compared to Agent V2, it is a major leap forward. It is 10x more autonomous, with the ability to periodically test your app in the browser and automatically fix issues using our proprietary testing system-3x faster and 10x more cost-effective than Computer Use models.https://blog.replit.com matt palmer @mattppal 5 of the biggest features in the new Replit Agent 3 135 5:58 AM • Sep 11, 2025 2Replit raised $250M led by Prysm, nearly tripling its valuation to $3B; its annualized revenue grew from $2.8M to $150M in the past year and it has 40M usersReplit hits $3B valuation on $150M annualized revenue | TechCrunchUnlike other fast-growing vibe coders, Replit has not been running to VCs every few months, though we'll see if that changes.https://techcrunch.comhttps://www.prnewswire.com/news-releases/replit-closes-250-million-in-funding-to-build-on-customer-momentum-302551540.html 3via Cyrus Li of Bytedance, a PDF of about Bytedance's latest text-to-image and image-editing model Seedream 4.0. Also sharing various links of providers who are offering API access to Seedream 4.0 or incorporating it in their product https://www.notion.so/yusuf-goolamabbas-53/Public-Files-to-be-linked-in-Paragraph-26b83c2ae303801b9e9cdbd3c065e81e?source=copy_link#26c83c2ae303804aba3df8e3a95c4795 Freepik @freepik Introducing Seedream 4.0 Banana jokes aside, this changes everything – Generate in 2k and 4k resolution – Up to 6 image references – Choose any aspect ratio Unlimited generations for Premium+ and Pro 933 12:29 AM • Sep 10, 2025 fal @FAL Seedream 4.0 is launching on fal day 0, #1 model for multimodal image generation and editing. With exclusive discounts. Text-to-image and image editing in one powerful model Natural language control with pixel-perfect precision Just $0.03 per generation! Volume 692 4:38 PM • Sep 9, 2025 Replicate @replicate ByteDance's latest image model Seedream 4 is now on Replicate - generate images up to 4096x4096 for just $0.03 per image - edit images with prompts - use up to 10 reference images, and even generate multiple views and poses from a single prompt replicate.com/bytedance/seed… 454 8:04 PM • Sep 9, 2025 WaveSpeedAI @wavespeed_ai Seedream 4.0 is now LIVE on WaveSpeedAI! Unleash the future of AI image generation with: Precision prompt editing Extreme fidelity & feature preservation Deep intent understanding Multi-image input & output Ultra-fast, ultra-HD rendering From concept 219 5:01 PM • Sep 9, 2025 ComfyUI @ComfyUI Seedream 4.0 Now Available in ComfyUI! Generate, edit, multiple outputs, all in one model! 4K resolution with lightning speed Natural language control Up to 9 outputs and 6 inputs in 1 generation Consistency across multiple generations 239 6:01 AM • Sep 11, 2025 4Response from Cyrus: love to see more developers or creators to leverage our technology. Ping me if you'd like to talk about official partnerships. Cheers Response from Emmanuel de Maistre Seedream 4.0 is super cool Cyrus - We've already integrated it in Scenario.comFriday 12th September 2025Poster: Florence Just discovered this, has anyone tried it out yet? It’s called “Oboe” https://oboe.fyi/news Oboe @oboelabs Introducing Oboe, the easiest way to learn anything, with magical courses made just for you. We’re heading toward a future where humans only exist to feed AI. AI gets smarter, we get stupider. But what if AI’s purpose was to feed us? That’s the future we hope Oboe can help nudge 740 10:00 PM • Sep 10, 2025 Saturday 13th September 2025Matthew Berman interviewing Amjad Masad, co-founder/CEO of Replit I welcome feedback not only on this newsletter but across all the various surfaces where I curate content Regards, Yusuf

## Publication Information

- [This Week in All Things AI](https://paragraph.com/@twiata/): Publication homepage
- [All Posts](https://paragraph.com/@twiata/): More posts from this publication
- [RSS Feed](https://api.paragraph.com/blogs/rss/@twiata): Subscribe to updates
- [Twitter](https://twitter.com/yusufg): Follow on Twitter