- Everyday AI
- Posts
- Ep 632: ChatGPT Apps: 3 Hands-on approaches to save time today
Ep 632: ChatGPT Apps: 3 Hands-on approaches to save time today
How to use ChatGPT apps, Google Gemini 3.0 Pro leaks, Mark Cuban warns against āeroticā ChatGPT mode, Google drops impressive AI video updates and more.
Everyday AI
October 15, 2025
š Subscribe Here | š£ Hire Us To Speak | š¤ Partner with Us | š¤ Grow with GenAI
Sup yāall š
Weird ask.
Can you reply to this email and let me know how we can make Everyday AI better? Seriously.
We work for you.
āļø
Jordan
Outsmart The Future
Today in Everyday AI
8 minute read
š Daily Podcast Episode: The future of work is Apps inside of LLMs. We give you a preview. Give it a watch/read/listen.
šµļøāāļø Fresh Finds: Anthropic released Haiku 4.5 for free, Excel debuts useful new AI commands, and experts say thereās āabsolutelyā an AI bubble. Read on for Fresh Finds.
š Byte Sized Daily AI News: Google Gemini 3.0 Pro leaks, Mark Cuban warns against āeroticā ChatGPT mode, Google drops impressive AI video updates and more. Read on for Byte Sized News.
šŖ Leverage AI: Not sure if ChatGPTās Apps will move the needle for you? Chances are, one of our 3 use-cases today will hit the spot. Keep reading for that!
ā©ļø Donāt miss out: Did you miss our last newsletter? Whatās happening: OpenAI says yes to erotica, Walmart and ChatGPT teaming up for AI shopping and Salesforce drops Agentforce 360 with OpenAI. Check it here!
Ep 632: ChatGPT Apps: 3 Hands-on Approaches to Save Time Today.
You haven't used ChatGPT's Apps yet? š«
Oh.... you like wasting time?
Even for free users, ChatGPT rolled out its new Apps mode that promises to shift the future of work.
Don't know how to work it? Don't know where to start?
Also on the pod today:
ā¢ File finishing friction solved? šļø
ā¢ Figma mind map magic š§
ā¢ Free users get apps now šø
Itāll be worth your 44 minutes:
Listen on our site:
Click to listen
Subscribe and listen on your favorite podcast platform
Listen on:
Hereās our favorite AI finds from across the web:
New AI Tool Spotlight ā Emergent is an AI app builder that ships full stack apps, Supercut is AI video messaging for teams, KaneAI is a GenAI native testing agent.
AI Dev Updates ā OpenAI released an updated version of GPT-5 for devs to filter the websites it visits in the API.
We shipped an updated web search model in Chat Completions: gpt-5-search-api.
š It's 60% cheaper at $10/1K calls and includes domain filtering, just like web search in Responses.
ā OpenAI Developers (@OpenAIDevs)
10:19 PM ā¢ Oct 14, 2025
AI Bubble Talk ā Wall Streetās getting jittery as AI stocks soar, with big names warning the boom is starting to look āabsolutelyā like a bubble.
AI Data Partnerships ā Nscale scored a $1.7B boost and a deal with Microsoft to deploy 200,000 Nvidia GPUs in new AI data centers.
Claude Updates ā Anthropicās Claude Haiku 4.5 is now free on Claude.ai and delivers near-frontier coding speed and quality.
Apple AI Chips ā Appleās just released M5 chip makes onādevice AI much faster across Mac, iPad, and Vision Pro.
Excel and AI ā Microsoft released new AI features in Excel.
This is super useful! With new Formula Completions in Excel, just type "=" and Copilot proactively suggests a formula, based on the context of your sheet. Here's a great example.
ā Satya Nadella (@satyanadella)
7:03 PM ā¢ Oct 14, 2025
1. Gemini 3.0 leak points to October 22 reveal š¤«
According to 9to5Google, a sketchy internal calendar screenshot suggests Google may announce Gemini 3.0 on October 22, although availability might lag the unveiling. The leak shows internal testing and bug checks underway, but the source image lacks a clear trail, so treat it with caution.
Googleās last major step was Gemini 2.5 Pro in March, which pushed āthinkingā and complex post-training, setting expectations that 3.0 could be a meaningful capability jump
2. Google drops Veo 3.1 for faster, smarter image-to-video magic š¹
Google DeepMind has released Veo 3.1, a video model that blends multiple images into cohesive clips, improves realism and audio, and adds features like first-to-last frame interpolation and scene extension.
The update arrives just months after Veo 3 and includes a lighter Veo 3.1 Fast, signaling a quickening pace in AI video tools that can turn a few stills into polished footage with richer sound and better storytelling. Available in Flow, Vertex AI, the Gemini API, Gemini App, and Vids, it marks a clear shift toward low-lift video production that small teams can actually use.
3. Mark Cuban warns OpenAIās erotica plan could trigger a parent-school revolt š¢
Billionaire Mark Cuban slammed Sam Altmanās plan to allow erotica in ChatGPT for āverified adultsā starting in December as reckless, arguing parents and schools will block the product if kids can bypass age gates, according to Fortune.
Altman says the shift will make ChatGPT feel less restrictive and closer to what people liked about 4.0, but critics point to mounting psychological risks for minors forming secret emotional bonds with chatbots and to lawsuits alleging AI-driven harm.
4. Google unveils 27B cell-model breakthrough ļæ½*
According to Google Research and Yale University, a new 27 billion parameter model called Cell2Sentence-Scale (C2S-Scale) uses Gemma to interpret single-cell biology and just generated a lab-confirmed hypothesis that could help turn ācoldā tumors āhot.ā
The model flagged the CK2 inhibitor silmitasertib as a conditional amplifier that boosts antigen presentation only when low interferon is present, a context-specific effect their smaller models missed and that saw roughly a 50% lift in vitro. This matters now because it suggests larger biological models are not just better at known tasks but can surface novel, testable ideas that could accelerate drug discovery and combo therapies.
5. Fresh antitrust suit challenges MicrosoftāOpenAI pricing power š¤
According to Computerworld, a new lawsuit filed in the Northern District of California by businesses alleging overcharges accuses Microsoft of using its OpenAI partnership to inflate ChatGPT prices by 100 to 200 times competitors during a February 2025 price war. Microsoft says the partnership promotes competition and innovation, while analysts note the case faces a high bar but could push transparency on AI pricing and compute access.
Regulators in the UK and EU previously declined to treat the relationship as a merger, and OpenAIās ability to source non-Microsoft compute may weaken claims of control, yet the filing seeks treble damages and casts pricing as a structural risk.
6. Gemma model helped uncover new cancer therapy pathway š§āāļø
Google and Yale just launched a massive 27-billion parameter AI model that can predict new ways to make cancer drugs work better, as reported on bioRxiv.
The model discovered a drug combo that helps the immune system spot hidden tumors, and lab tests backed it up. This breakthrough shows bigger AI in biology can directly lead to life-saving therapies, not just smarter predictions.
š¦¾How You Can Leverage:
Mountains of brilliant AI-generated content are dying inside chat windows because nobody wants to deal with the copy-paste-reformat nightmare.
Real talk shorties.
We've all done it. Generate something chef kiss perfect in ChatGPT, then spend 20 minutes fighting with formatting in another app.
WTF is that gray outline?!
Or worse, you just use an inferior AI integration that's already baked into a SaaS tool you're using because the extra steps ain't worth it.
We broke down this file finishing friction problem on today's episode of Everyday AI, and the solution just dropped from OpenAI.
(Well, maybe, if they don't screw it up)
Seven new apps let you natively use apps with your words and finish deliverables inside ChatGPT without ever leaving the conversation window.
No transfer tax. No format fallout. Just getting work done.
Time to capitalize. Lez goooo.
Make sure to go check out today's show, but here's the 1-2-3 of what ya need to know.
1 ā Computer Vision Rebuilds Complex Visuals in Seconds š„
Upload a diagram and watch ChatGPT apps analyze structure, rebuild every component, and apply your changes simultaneously.
Wild.
We tested this with a mind map that would've taken 5 hours to recreate manually in Figma. You know the drill. Click a shape. Duplicate it 47 times. Draw lines between them. Watch your soul leave your body.
Nah.
Flipped the orientation from horizontal to vertical, changed the entire color scheme, and had it done in 90 seconds. Then told it to consolidate 4 separate categories into 2 using logical reasoning.
It combined related nodes while preserving all underlying data.
The breakthrough is computer vision working as active context inside professional design tools. The app doesn't just import your visual and call it a day. It understands the structure, relationships between elements, and hierarchy. Then it applies complex transformations you describe conversationally.
No googling "how to rotate layout in Figma" at 11 PM.
Try This
Find your most complex workflow diagram buried in old files and upload it to ChatGPT with Figma connected. Describe all the structural changes you've been avoiding in one prompt like you're texting a designer. Watch it rebuild everything while applying changes simultaneously, then iterate using conversational commands. The output stays fully editable in Figma at 80% completion so your designers refine from there instead of starting from zero.
2 ā Inline Learning Support Kills Context Switching ā”
Educational content and AI assistance finally live in the same window that understands how you actually learn.
This is huge fam.
Tell ChatGPT your learning style before launching a Coursera course and it personalizes every explanation to match your brain. We mentioned learning best through 90s basketball references, then started a neural networks course directly in the chat window.
Paused mid-lecture to ask about deep learning. Got an explanation using Chicago Bulls analogies that connected to earlier context automatically.
No switching tabs. No losing your place. No frantically scribbling notes across 3 different apps while rewinding the same 30-second clip.
OpenAI's internal testing showed 70% productivity increases. This is exactly why.
The app remembers your professional background, knowledge gaps, and how you process information while you're consuming course content. You're not just passively watching videos anymore hoping something sticks.
You're having a conversation with material that adapts to your context in real time.
Try This
Before starting your next training course, give ChatGPT 90 seconds of context about how you learn and what you already know. Connect Coursera through the apps menu and launch content directly in the chat window. Pause whenever concepts feel unclear and ask questions like you're talking to a tutor. The AI references both the course material and your personal context to explain things in ways that stick for you specifically.
3 ā Enterprise Data Becomes Creation Context Automatically š
The biggest productivity killer is manually bridging the gap between your research and creation tools.
Run deep research through connected enterprise data before launching apps and context flows automatically. We analyzed an entire month of work documents using ChatGPT connectors and extracted 20 specific trends across hundreds of pages.
Then immediately asked the Canva app to generate a 5-page visual guide.
No copying insights into a Google Doc first. No reformatting research 47 times. No switching between 12 browser tabs while trying to remember which stat came from which source.
The connector analyzed 29 different documents, spotted patterns across unrelated content, and handed everything to the app as enriched starting material in one shot.
Canva's internal stats showed 400% faster design completion using this workflow versus traditional methods. That's the difference between spending your entire afternoon on a presentation versus finishing it in 20 minutes while maintaining higher quality.
(Your boss is gonna think you hired a designer)
The apps pull from your actual enterprise data sources to understand previous work patterns, communication style, and content preferences. This compound effect separates teams who've solved the file finishing friction problem from everyone else still manually copying information between programs like it's 2019.
Try This
Connect your work tools through ChatGPT connectors right now and run a deep research query on your recent projects. Ask it to identify patterns or spot trends across everything you've created in the last month, then let it analyze for 10-15 minutes. Immediately launch the relevant app and request deliverables using only that research as context. Test this Monday morning on your weekly status update before your team figures out why yours suddenly look 5 times better.
Reply