Anthropic's most powerful model is here

Welcome back, Superhuman. Anthropic dropped Claude Opus 4.5 yesterday, declaring it the “world's best coding model.” OpenAI chimed in by turning ChatGPT into your personal holiday shopper, while Harvard released a disease-detecting AI. Quite the Monday — let’s break it down.

Today: Anthropic’s new coding powerhouse, learn how to automate your email and calendar with agents, and get the latest prompts and trending social posts.

TODAY IN AI

1. Anthropic releases Claude Opus 4.5: Anthropic’s latest model significantly outperforms competitors at coding, agentic tasks, and computer use, while also being more affordable than Claude Opus 4.1. The AI leader also announced updates to Claude Code, Claude in Excel/Chrome, and the Claude Developer Platform. Opus 4.5 is currently available — See it in action here.

2. OpenAI launches free “shopping research” feature for ChatGPT: ChatGPT can now help you shop for products by asking you clarifying questions, researching the web, and bringing options back to you as it refines the results. The feature is rolling out to all users now, with nearly unlimited usage available to all plans through the holidays. See how it works here.

3. AI model PopEVE enhances diagnosis of rare genetic diseases: Developed by Harvard Medical School researchers, PopEVE excels at identifying genetic variants most likely to cause disease — outperforming competitors like Google DeepMind's AlphaMissense. Harvard’s team is now testing the model in clinical settings to determine if it can accurately diagnose rare, single-variant genetic conditions.

PRESENTED BY SANA

Your AI operating system for real work

Sana Agents isn't another AI tool. It's your AI operating system.

Deploy no-code agents that handle your entire workflow. They autonomously search your Drive for pricing docs, pull context from your CRM, analyze data for dashboards, and generate presentations—across 100+ applications. Powered by A-5, our proprietary multi-agent system that plans, acts, and reflects autonomously.

The results speak for themselves. Fortune 500 teams save 10 hours per week per employee, cut prep time by 62%, and complete compliance audits 66% faster.

See how it works

THE AI ACADEMY

How to create AI agents that automate your email, schedule your meetings, and more (in 16 mins)

AI agents are getting good — really good. We’ve gone from simple agents like ChatGPT Deep Research to more complex agents that can handle tasks across multiple apps like your email and browser.

I got Flo Crivello (CEO of agent platform Lindy) to show you how to build agents in just a few clicks that will:

• Automate your inbox and respond to your emails

• Schedule meetings without any input from your end

• Prepare meeting notes for your calls by searching the web

• Automate browser based tasks like data copy pasting across apps

👉 Watch the full tutorial here

FROM THE FRONTIER

AI with human-level intelligence is coming. But experts are split on the arrival date.

Artificial general intelligence (AGI) is on its way, but not this year. In January, prediction platform Polymarket gave OpenAI a 35% chance of announcing AGI in 2025 — fairly good odds for such an immense breakthrough. Since then, that figure has dropped to just 2%. So what changed?

GPT-5's August launch didn't help. ChatGPT-5 was widely expected to be a landmark breakthrough that would inch us closer to AGI, but it largely fell short of those expectations. Instead of a pivotal leap forward, GPT-5 felt like an incremental update — prompting one prominent critic to call it "overdue, overhyped and underwhelming." Still, those who work in the industry largely agree AGI remains on the roadmap.

Experts still expect AGI, they just can't agree when. The AI 2027 report, led by ex-OpenAI researcher Daniel Kokotajlo, originally forecasted AGI by 2027 but has since pushed the timeline to around 2030. Metaculus, a well-regarded forecasting platform, projects August 2033. The range is wide, but the direction is clear: it's coming.

Here’s the bigger picture. Just four years ago, Metaculus forecasters predicted AGI wouldn't arrive for another 50 years. Now most experts expect it within a few years. The window has collapsed dramatically, and with it, the time society has to prepare.

PRESENTED BY SEMRUSH

Your business, always visible.

The AI agent that runs your Google Business Profile — while you focus on growth.

• Up to 40% visibility increase in local search

• 5-minute setup, continuous optimization

• No expertise needed

Activate AI Agent

IN THE KNOW

Today’s trending social posts and use cases

📺️ YouTube Graphics: Learn how to convert entire YouTube videos into detailed infographics — just by copying the video’s link.

🤳 Creative Prompt: One clever AI trick? Generating “behind the scenes” shots of existing photos. Learn the prompt here.

📝 Show Your Work: Gemini can now solve math problems directly on a worksheet — and even match your handwriting in the process.

😸 Cat Dough: The latest viral AI video? Cats being kneaded like dough (It’s got 1.1 million views).

🐲 Dragon Wash: Watch Toothless from How to Train Your Dragon enjoy a post-flight shower.

PRODUCTIVITY

5 New & Trending AI Tools

• 🤖 UiPath*: Meet UiPath Maestro, the orchestration engine behind one of TIME’s Best Inventions of 2025. Bring AI agents, robots and people together in one secure flow.

• 📱 UX Magic: Turn simple prompts into complete, production-ready UI and wireframe designs in seconds.

• 🧑‍🏫 Guideflow: Create interactive demos and guides in seconds with AI.

• 🤝 ReadMeeting: Summarize and record meetings in real-time.

• ⚡️ Browser Cash: Browser automation platform for AI agents, web scraping, and internet intelligence.

^{* indicates a promoted tool, if any}

PROMPT STATION

The 360° Decision Prompt

Prompt: I’m stuck on [problem/decision].
Break down my situation using the following structure:

1. Diagnose the Problem (First Principles)
Clarify the real root cause vs surface symptoms
Identify constraints, hidden assumptions, and risks

2. Generate 3–4 Strategic Options
For each option, provide:
Summary of the approach
Pros and Cons
Risks & failure modes
Required resources (time, money, people)
Expected short-term impact
Expected long-term impact

3. Decision Matrix (Score Each Option 1–10)
Use a weighted scoring model that includes:
Impact
Effort required
Reversibility (how easy it is to undo)
Speed of execution
Strategic alignment
Risk level
Then calculate a Total Score and rank the options.

4. Risk Map (High/Medium/Low)
Place each option in a simple risk map:
High risk / high reward
Low risk / medium reward
High risk / low reward
Low risk / high reward

5. Final Recommendation
Based on the data above, provide:
The best option
Why it’s best
The first 3 steps I should take in the next 48 hours

6. Bonus: If you see a better option I didn’t consider, propose it.

Keep your tone objective, strategic, and practical.

Holiday Fashion

Midjourney Prompt: custom [object/product] designed in a [theme/style] theme. The [object/product] should have [specific decorative elements], such as [example elements]. The resolution should be hyper realistic and show details in the materials being used.

(Pro Tip: You can use the same prompt in ChatGPT to get similar results)

Whenever you’re ready to take the next step

• Check out our Top 125 AI Tools

• Get better at prompting with our Top 1,000 Prompts

• Learn how to use AI at work and get certified with our free course

Grow customers & revenue: Join companies like Amazon, Hubspot, and Salesforce. Showcase your product to our 1M+ readers and 2M+ followers on socials. Get in touch.

Until next time — Zain & the Superhuman AI team