AI in a Nutshell - Week 48 - Claude Has Better Context Window, Sora Gets Rate-Limited, and Openai Fires Mixpanel
Anthropic now has better context window, Amazon commits $50B for government AI infrastructure, OpenAI integrates voice directly into ChatGPT, and Sora gets rate limited.
Fellow human, it’s week 48. AI hasn’t taken over the world yet. And here’s your lazy-man version of what went down in the AI world in the past week.
Summary for extra lazy people: Anthropic claims Claude boosts productivity by 80% and now has a better context window, Amazon commits $50B for government AI infrastructure, OpenAI integrates voice directly into ChatGPT, Sora gets rate-limited, and both OpenAI and Perplexity launch AI shopping assistants.
Quick Note: I’m not an expert, I’m just a curious normie navigating the world of AI (like you). Every Sunday morning, I wake up super early (before the kid wakes up), go online, and read about AI, like my dad used to read newspapers back in the day. Then I write down everything new I discovered and turn it into this newsletter for anyone else who’s curious enough to read it. Why? Because I learn faster and understand deeper when I have to explain it in writing. I’m welcome.
What You Must Know
Anthropic Estimates Productivity Gains from Claude → Anthropic released a research analyzing time savings from Claude conversations, they estimated an average 80% speedup on tasks. I’m always skeptical when companies measure their own product’s impact; it’s a “judge and jury” situation 😀, but if even half of this is true, it’s a big deal.
OpenAI and Perplexity Roll Out AI Shopping Assistants → I saw on Techcrunch that both OpenAI and Perplexity launched AI shopping features inside their chatbots to help users research and compare purchases. I believe “shopping” is an everyday use case, and I previously wrote an opinion piece about why I think AI Agent for everyday use is “not there” yet, but I’m hopeful to see how using AI for shopping will evolve over time.
OpenAI Integrates ChatGPT Voice → OpenAI has integrated ChatGPT Voice directly into the main chat interface, so we no longer need to use a separate mode. That means we can now talk, view responses, and access earlier messages in real time.
Claude Can Now Have Longer Context Windows → Everyoneeeee hates Claude’s context window limit. But now, Anthropic has updated Claude to auto-compact older messages and prevent context limit interruptions mid-chat. This means you can have longer conversations without hitting the dreaded “context window full” wall.
OpenAI Suffers Data Breach Through Third-Party Mixpanel → Within the week, I received several emails about the Mixpanel (an analytics tool) breach. But OpenAI’s was the most worrisome. In their clarification, they mentioned that only basic profile information was exposed. Not a major breach, but still embarrassing for a company that handles so much sensitive data. Now, word on the street is that OpenAI has dumped Mixpanel.
What’s Good to Know
Alphabet Nears $4T Valuation on AI Optimism → Alphabet’s (the parent holding company of Google) market value is on track to hit around $4 trillion due to the market pricing Google as one of the main long-term winners of the AI race.
Jony Ive and Sam Altman Say They Finally Have an AI Hardware Prototype → I read on The Verge that Jony Ive and Sam Altman now have a working prototype for OpenAI’s first hardware device, which Ive mentioned will potentially arrive in “less than” two years. The details are still vague, and I'm not sure if it’ll be a playful AI gadget or a phone-shaped device. Given Ive’s design legacy, people are expecting something interesting.
Sora Gets Rate-Limited from High Demand → OpenAI throttled generation limits on Sora, mentioning huge demand over the holiday period. In simple terms, people are hammering these models so hard that the company had to hit the brakes, which says a lot about current AI usage levels.
Character.AI Launches “Stories” Mode for Teens After Banning Open Chat → After announcing it would cut off open-ended chats for teens, Character.AI rolled out a new “Stories” feature pitched as a more structured, safer way for under-18s to interact with its AI. Teens get guided, story-like experiences instead of free-form conversations. I’d say it’s better, in my opinion.
Amazon Plans Up to $50B for US Government AI Supercomputing → Amazon said it could invest up to $50 billion building high-security AI and high-performance computing data centers for U.S. government agencies.
Money Moves This Week
💰 Onton raised $7.5M to expand its AI shopping platform.
💰 Model ML raised $75M to automate junior bankers’ work with AI.
💰 Uxia raised €1 million in pre-seed funding to develop its AI software and SaaS platform.
💰 AskCipher raised an undisclosed amount in strategic funding to advance its AI and SaaS solutions.
💰 Ranketta raised €1 million in pre-seed funding to build its AI and SaaS platform.
💰 Bonus Find → 49 US AI startups have already raised $100M+ in 2025
Kaizen Corner → Lenny’s Product Pass Might Be Worth Checking Out
I found something on Lenny’s Newsletter that caught my attention: Product Pass.
It’s basically a bundle subscription that gives you access to a bunch of product management and tech tools in one package.
For an annual subscription to Lenny’s Newsletter (around $200-350/year), you can get a free year of 15+ products, including things like Devin, Superhuman, Linear, Perplexity, Lovable, and more. The total value is supposedly over $10,000.
I didn’t need it because I already have subscriptions to most of them, but if you use a lot of these tools or want to try them out, this could be a solid deal.
Note: Some products might run out of codes since they’re limited, and you need an annual subscription (not monthly) to access the bundle.
Check it out here if you’re interested.
Meme of the Week
That’s your week in AI.
If you learned something, tell a friend. And if you didn’t, blame yourself :)
Until next Sunday,
Kay - your fellow human.
P.S. If this email lands in spam, that’s your inbox trying to stop you from staying plugged in. Fix it.