Claude to rectify mistakes 4x faster

Welcome back to The Shift. Let’s get straight to what matters in AI today…

Today we have:

🤖 Anthropic just released Claude Opus 4.8, and It Is Four Times Less Likely to Miss Its Own Mistakes

🧠 How to build an autonomous research agent with NotebookLM + Google Antigravity

📱 Apple is rebuilding Siri from the ground up on Google Gemini for iOS 27

🔨Tools and Shifts you cannot miss

🤖 Anthropic Just Released Claude Opus 4.8 and It Is Four Times Less Likely to Miss Its Own Mistakes

Anthropic released Claude Opus 4.8 at the same price as its predecessor, alongside new effort controls, a major Claude Code upgrade, and a $65B Series H that puts its valuation beyond OpenAI at $965B.

The Shift:

Opus 4.8 Beats GPT-5.5 and Gemini 3.1 Pro on Key Benchmarks - It outperforms both on agentic coding, computer use, financial analysis, and Humanity's Last Exam, coming in at the same price as Opus 4.7.

It Flags Its Own Errors Instead of Glossing Over Them - Opus 4.8 is four times less likely than Opus 4.7 to let code flaws pass unremarked, flagging uncertainties more often instead of making confident, unsupported claims.

Effort Controls and Fast Mode Give You More Flexibility - A new effort control on claude.ai lets you choose how hard Claude thinks. Fast mode is now three times cheaper at 2.5x speed across all plans.

Claude Code Can Now Run Hundreds of Parallel Subagents at Once - Dynamic workflows let Claude plan large jobs, split them across parallel subagents, verify outputs, and complete codebase-scale migrations in a single session.

Alignment scores improved significantly with lower rates of deception than Opus 4.7. Mythos-class models are coming in the next few weeks, currently in limited preview for cybersecurity work via Project Glasswing.

Together with Wispr Flow

Talk to your AI tools the way you'd talk to a colleague.

You don't send a colleague a three-word brief. You explain the context, the constraints, what you've already tried. But typing all that into ChatGPT takes forever — so you don't.

Wispr Flow lets you speak your prompts instead. Talk through your thinking naturally and get clean, paste-ready text. No filler words. No cleanup. Just detailed prompts that actually get you useful answers on the first try.

Millions of users worldwide. Works system-wide on Mac, Windows, and iPhone.

Try Wispr Flow free

🧠 How to Build an Autonomous Research Agent With NotebookLM + Google Antigravity

Research usually breaks because you collect sources in one place, analyze them somewhere else, then manually turn them into notes. This workflow lets Antigravity do the heavy lifting while NotebookLM becomes the organized research brain.

Step 1: Create one NotebookLM project - Open NotebookLM and create a new notebook around one big topic, like “AI shopping agents,” “DTC retention trends,” or “competitor research.”

Step 2: Open Google Antigravity - Start a new chat or workspace in Antigravity and define the research goal clearly.

Step 3: Give it a research brief - Prompt it with the topic, target audience, research depth, and exact outputs you want, such as summaries, insights, examples, risks, and opportunities.

Step 4: Ask it to gather sources - Tell Antigravity to find credible articles, reports, transcripts, docs, and data points, then organize them by subtopic.

Step 5: Push sources into NotebookLM - Have Antigravity add each source into the right NotebookLM section so the notebook becomes a clean research base.

Step 6: Generate the final insight layer - Ask NotebookLM to summarize patterns, compare sources, extract key claims, and turn the research into an article outline, report, or strategy brief.

Once set up, you get a repeatable research engine, Antigravity collects and organizes, NotebookLM reasons over the sources, and you turn complex topics into usable outputs much faster. You can try here: NotebookLM | Google Antigravity

📱 Apple Is Rebuilding Siri From the Ground Up on Google Gemini for iOS 27

Apple is overhauling Siri for iOS 27, moving it into Dynamic Island and rebuilding it on Google Gemini. Bloomberg got an early look at what is shaping up to be the most significant Siri redesign since the assistant launched in 2011.

The Shift:

Siri is moving into dynamic island with a swipe-down interface - A swipe down opens an AI interface that handles search, chat, and iOS tasks using on-device data, screen content, and the web, all without leaving what you are doing.

It runs AI web search like Perplexity and surfaces answers as rich cards - Rebuilt on Google Gemini, Siri will run AI-powered web searches with answers appearing directly inside Dynamic Island as structured rich cards rather than links.

Swiping further opens a dedicated ChatGPT-style Siri app - Users can drop into a full conversational AI interface and choose to route queries to external AI models, giving them flexibility beyond what Apple or Gemini provides natively.

Siri is also coming to the camera app with advanced AI editing - AI-based photo editing, AI wallpapers, and natural language shortcut creation are all in the pipeline, expanding Siri beyond search and into creative and system-level tasks.

This is Apple's clearest signal yet that it is leaning on external AI partnerships rather than building everything in-house. Routing queries to third-party models directly inside Siri changes what the assistant fundamentally is.

Together with Pulley.com

Cap table tools founders and finance teams love

As you grow, you need a cap table management platform that you can trust to keep up. 

From issuing grants to 409A valuations or ASC 718 reporting, Pulley gives you the tools to manage equity. Experience a platform built for you: transparent, reliable, and designed to put your company first.

Learn why companies choose Pulley

🔨AI Tools for the Shift

⚽ Hobbit AI – Helps youth football teams with AI-powered coaching insights, training support, and player development assistance.

🎤 Lip Sync AI – Creates perfectly synced lip-sync videos from photos, videos, and audio clips in seconds with no signup required.

🧑‍💼 AvatarCraft AI – Generates professional AI avatar videos from text or audio for marketing, training, and social content.

🖋️ Tat Ink – Instantly creates custom tattoo designs in multiple styles with high-resolution artwork ready for printing or reference.

💬 Daysi AI – Captures customer enquiries, answers questions instantly, and identifies what is hurting your conversion rates.

🚀Quick Shifts

🎨 Adobe’s Firefly AI Assistant is a conversational design helper that can perform image-editing tasks through natural-language prompts. Early tests suggest it reduces busywork while keeping creators involved, though its capabilities remain limited. 

⚡ Microsoft launched a redesigned Microsoft 365 Copilot that it says loads 2x faster, delivers more structured responses, introduces adaptive “progressive disclosure” controls, and adds richer in-app AI interactions across Word, Excel, and PowerPoint. 

🎙️ YouTube is adding podcast-focused features for Premium users, including an audio-first “On-the-Go” mode, AI-powered automatic speed adjustments, and chatbot-based podcast discovery, bringing it closer to dedicated podcast apps like Spotify. 

🤖 Asana is acquiring no-code AI agent builder StackAI for $75 million, accelerating its AI strategy and expanding automation capabilities for complex enterprise workflows and human-agent collaboration.

💳 Meta is rolling out paid subscriptions across its apps, including Instagram Plus ($3.99/month), Facebook Plus ($3.99/month), and WhatsApp Plus ($2.99/month), while also testing premium creator, business, and AI-focused plans under the new Meta One brand. 

🧩 Prompt of the Day

AI & Automation Opportunities That Actually Save Time and Scale Growth

Most businesses use AI for isolated tasks. The real advantage comes from integrating automation into workflows that improve speed, decision-making, and customer experience.

Turn AI into an operational multiplier, not just a trend.

Paste the prompt: Drop this into ChatGPT, then fill in your business area.

Prompt to paste

Create an AI and automation strategy for [Insert business area or department]. Identify high-impact use cases where AI or automation can improve workflows, reduce manual effort, increase speed, or enhance customer experience. Explain the expected benefits such as cost savings, faster execution, personalization, better decision-making, or higher engagement. Outline implementation steps including recommended tools, integrations, workflow changes, team responsibilities, and rollout priorities. Focus on practical opportunities that deliver measurable operational and business improvements.

🤳AI Nugget of the Day

That’s all for today’s edition. See you tomorrow as we track down and get you all that matters in the daily AI Shift!

If you loved this edition, let us know how much:

Forward it to your pal to give them a daily dose of the shift so they can 👇