This website uses cookies

Read our Privacy policy and Terms of use for more information.

Presented by

Good morning, AI enthusiasts. Apple just agreed to pay $250 million to iPhone owners who felt misled by Apple Intelligence, the AI suite it marketed heavily before many features were ready.

The case may be the clearest signal yet that "AI-powered" is no longer free advertising. If the features don't arrive, the bill does.

In today's recap:

  • Apple's $250M settlement over AI marketing claims

  • Anthropic launches 10 finance agents for enterprise

  • Organize complex topics fast with NotebookLM Mind Maps

  • Astrocade raises $56M to let anyone build games

  • 4 new AI tools, prompts, and more

APPLE

Apple's AI settlement sets a precedent for the industry

Recaply: Apple just settled a $250M lawsuit over claims it oversold Apple Intelligence, with eligible iPhone owners set to receive $25 to $95 each in the first major AI hype accountability case.

Key details:

  • The lawsuit claimed Apple marketed AI features it couldn't deliver. Buyers upgraded their iPhones for tools that arrived late or not at all.

  • The $250M covers millions of eligible iPhone owners. Individual payments range from $25 to $95, based on the model purchased and claims filed.

  • Clarkson Law Firm led the case as co-lead class counsel, representing buyers who chose their iPhone based on Apple's AI marketing promises.

  • The settlement is in preliminary approval. iPhone owners can file claims as it moves toward a final court decision.

Why it matters: Apple spent the last year defending its AI promises. This $250M settlement says those promises cost too much. Apple launched Apple Intelligence with features that were delayed or never fully arrived. For every tech company putting "AI-powered" on every product label, this sets a new rule: deliver what you sell. The bill for AI hype has arrived, and it won't be the last.

PRESENTED BY CLICO

Every open tab becomes context for your AI

Recaply: Most AI tools only see what you paste into them. Clico reads across all your open tabs and lets you pull any of them into your prompt with @, so you can reference a Notion doc, a Gmail thread, and a spreadsheet without leaving the page you're on.

With Clico, you get:

  • @ to pull any open tab into your AI prompt instantly

  • Cross-tab context without switching windows or copying text

  • Full page or PDF summaries on a double-press

  • Memory that references past conversations as context

ANTHROPIC

Anthropic's finance agents take on OpenAI's excel play

Recaply: Anthropic just launched ten ready-to-run finance agent templates with Microsoft 365 integration, making a direct enterprise play against OpenAI's Excel add-on announced on the same day.

Key details:

  • Each template packages skills, connectors, and subagents into a reference architecture. The agents cover tasks from pitchbook building and KYC screening to month-end closing and ledger reconciliation.

  • Claude Opus 4.7 powers the templates and leads the Vals Finance Agent benchmark at 64.37%. The 10 agents span research, client coverage, and financial operations.

  • Claude carries context automatically across Excel, PowerPoint, Word, and Outlook. Work that starts in Excel doesn't need to be re-explained in PowerPoint, per Anthropic's announcement.

  • Available now as plugins in Claude Cowork and Claude Code. Outlook support is coming soon, with Claude Managed Agent cookbooks ready to deploy today.

Why it matters: OpenAI announced a ChatGPT add-on for Excel and Google Sheets the same day. But Anthropic isn't playing in the spreadsheet lane. Ten task-specific finance agents, context that flows across every Microsoft app, and a benchmark lead at 64.37% is a different level of enterprise ambition. This is Anthropic building for firms that want AI running the full deal workflow, not just one spreadsheet tab.

GUIDES

Organize complex topics fast with NotebookLM Mind Maps

Recaply: In this tutorial, you will learn how to use NotebookLM's new Mind Map feature to visually organize any complex topic, customize the map to your focus, and share it with collaborators without leaving your research.

Step-by-step:

  1. Go to notebooklm.google.com and create a new notebook. Upload your sources on the topic you want to map. PDFs, Google Docs, web URLs, and YouTube videos all work.

  2. Click "Mind Map" in the left panel. The tool builds a visual map of key concepts and connections extracted from all your uploaded sources.

  3. Click any node to rename it with your own label. Drag branches to reorganize the layout. Delete nodes that don't fit your focus area.

  4. Use the scope selector to filter the map to one source or one topic cluster. This lets you drill into a single document or zoom out across everything you've uploaded.

  5. Click "Share" and copy the link. Send it to teammates, paste it into meeting notes, or use it as a visual summary for a study session or presentation.

Pro tip: Upload 3-5 sources that cover the same topic from different angles. A research paper, a Reddit thread, and a YouTube explainer work well together. Let the Mind Map show you where sources agree and where they diverge.

TOGETHER WITH ATTIO

GTM Atlas, by Attio

GTM Atlas is a free resource every operator should read. Curated by Attio, the AI CRM, and written by GTM leaders from Lovable, Granola, and Vercel, you'll get:

  • ICP, outbound, and retention frameworks from operators who've built them

  • The qualification signals that actually predict conversion

  • Conversion plays that don't rely on a pitch deck

Mapped by operators. Curated by Attio.

ASTROCADE

Sequoia backs Astrocade's $56M bet on AI Game Creation

Recaply: Astrocade just raised $56M in a Sequoia-led round to scale its AI-native game creation platform, which reached 20 million users in just eight months with no coding required.

Key details:

  • Astrocade lets anyone turn a game idea into a playable experience in minutes using generative AI. No coding is needed, and creators can publish directly on the platform.

  • The platform reached 20 million users in eight months, with hundreds of millions of plays per month. Top creators are earning thousands of dollars monthly.

  • Founders Amir and Ali Sadeghian are brothers: a Stanford PhD and a former Google AI researcher with no gaming industry background. Investors include Sequoia, Sea, Google AI Futures Fund (the fund run by AI pioneer Fei-Fei Li), and NVIDIA.

  • The company launched publicly in late 2025. It plans to expand its team and deepen creator tools with the new funding.

Why it matters: Game creation has always required a studio, a budget, or years of training. Astrocade makes that feel like a distant memory. Its fastest-growing users aren't developers or designers. They're outsiders who've never written a line of code. When AI removes the production barrier, the next entertainment category won't come from game studios. It'll come from teachers, teenagers, and anyone with a weird idea worth playing.

TOOLS

Trending AI Tools

  • 📊 SubQ - Subquadratic's AI reasoning model with 12M-token context windows

  • 🔊 Realtime TTS-2 - Inworld AI's voice model for realtime conversation

  • 🤖 WorldRouter - WorldClawAI's AI agent routing layer for directing requests across models and tools

  • 💻 Sai - Simular AI's always-on AI coworker that handles computer tasks in the background

NEWS

What Matters in AI Right Now?

  • OpenAI rolled out GPT-5.5 Instant as a ChatGPT update, with the new version described as "smarter and more accurate" and featuring fewer unnecessary emoji in responses.

  • OpenAI launched ChatGPT for Excel and Google Sheets as a global add-on, letting users build, update, and explain spreadsheets directly within both platforms without switching apps.

  • Microsoft, Google, and xAI agreed to give the US government early access to new AI models for national security testing before public release, with CAISI reporting it has completed over 40 model evaluations including unreleased frontier models.

  • Google DeepMind UK workers voted to unionize and filed for recognition of two unions, citing concerns about the company's classified AI deal with the US military. At least 1,000 workers would be covered if Google grants recognition.

  • Meta is building a consumer version of OpenAI's OpenClaw agent internally codenamed "Hatch," while SAP moved to block OpenClaw and other unauthorized AI agents from its systems over licensing and product protection concerns.

  • Pennsylvania sued Character AI, claiming its chatbots illegally present themselves as licensed doctors and give medical advice to users, in what the Shapiro administration called a deceptive and dangerous practice.

  • Google released Multi-Token Prediction drafters for the full Gemma 4 family, delivering up to a 3x inference speedup via speculative decoding with no quality degradation, available now on vLLM, Hugging Face Transformers, and other frameworks.

EVENTS

🧡 Enjoyed this issue?

🤝 Recommend our newsletter or leave a feedback.

How'd you like today's newsletter?

Your feedback helps me create better emails for you!

Login or Subscribe to participate

Cheers, Jason

Connect on LinkedIn, & Twitter.

Reply

Avatar

or to participate

Keep Reading