Google teaches Gemini to daydream

In partnership with:

Hi AI Futurists,

Google’s new AI model, Gemini Deep Think, runs multiple lines of reasoning in parallel, allowing it to evaluate ideas simultaneously before committing to a response. It’s being deployed across Google tools, and early benchmarks show improvements in complex tasks like coding, planning, and writing. Let’s take a look.

Here’s our agenda.

• Our Sponsor: Mindvalley

• Top AI news

• Google Gemini’s Deep Think

• Top 3 selected AI tools

• AI in the wild

Best,
Lex Sokolin

If you're new to this newsletter, welcome! You're here due to your interest in AI and tech tools, and past engagement with similar content. Keep your subscription active to receive premium AI updates and expert tool recommendations. Manage your preferences below.

AI Mastery 2025

You may not know this yet — but Mindvalley is one of the most influential personal growth platforms in the world.

With over 20 million students, partnerships with Google, Harvard, and Forbes-ranked experts, Mindvalley is known for one thing:

Creating transformational experiences that help people thrive in the future.

And today, they’ve opened access to something that’s never been more timely:

A 22-week guided experience to help your audience lead (not lag) in the AI-powered future.

 And as of this morning, only 160 seats remain.

 Why This Is Blowing Up:

Since launching at the AI Summit last week, over 840 people have joined — from CEOs to solopreneurs to creators and career professionals.

They’re all saying the same thing:

“AI isn’t optional anymore — it’s the edge.”

This isn’t just a course. It’s a complete skillset upgrade that shows anyone how to:

✅ Reclaim 10+ hours/week with AI automation
✅ Create blogs, emails & videos 10x faster
✅ Launch AI workflows — without code
✅ Build a smarter business using today’s most powerful AI tools
✅ Earn a Mindvalley + CPD Certification to stand out globally

And they do it all in just 2–3 hours/week, with world-class support and live mentorship.

What Participants Are Saying:

“The bonus workshop with Vishen sealed the deal for me. This isn’t just theory — it’s real, actionable AI training.”
— Program Participant

“I had no idea I could automate so much in my business. I wish I started earlier.”
— Early Student

“This gave me the confidence to finally start using AI tools that used to intimidate me.”
— New Creator

Who This Is Perfect For:

• Coaches, creators & consultants who want to scale smarter

• Entrepreneurs and team leaders who want to future-proof operations

• Anyone who’s felt behind and needs a guided way in to AI

They also get a 15-day risk-free trial — so they can explore without pressure.

 Give Your Audience Access Before It’s Gone

Top AI news

🧠 AI Is Coming for the Consultants: McKinsey’s Existential Pivot
Consulting’s biggest talent incubator is shrinking staff and building bots instead.

📰  Google’s AI Overviews Spark Publisher Panic at NPR
Publishers warn AI summaries are decimating traffic nearly one click at a time.

🛠️ Hard‑Tech AI Is Silicon Valley’s New Fixation
Silicon Valley is shifting from selling software to building infrastructure.

🔍 Universal AI Detector Flags Deepfakes Even Without Faces
UNITE—a new model from UC Riverside and Google—sniffs out doctored videos by scanning full frames.

💼 The AI Super‑Stimulant That’s Powering U.S. Growth
Corporate spending on AI infrastructure is now the main driver of U.S. growth.

🧨 Vogue Sparks Outrage with AI‑Only Guess Ad
Readers called it lazy, cheap, and humanity‑free.

🚀 Apple's Whispered AI Coup: Building Its Own Answer Engine
Apple quietly launched an internal “Answers” team to build a ChatGPT-style engine, signaling a bold new assault on search dominance.

🧗 AI‑Powered Drone Finds Missing Hiker by Spotting Their Helmet
AI scanned 2,600 drone images and flagged a tiny red helmet in hours.

📡 Anthropic cuts OpenAI from Claude API
OpenAI lost API access to Anthropic’s Claude on August 1, 2025 for using Claude Code ahead of its GPT‑5 launch.

🧭 AI Won’t Save the Global South Unless Power Shifts
AI risks reinforcing colonial patterns unless local stakeholders own the tools.

Google teaches Gemini to daydream

Google has launched Gemini Deep Think, a new AI reasoning engine that tests multiple hypotheses at once. Unlike traditional models that complete one chain of thought before starting another, Deep Think evaluates a range of possibilities in parallel and then picks the most promising path. It’s a shift from sequential to branching logic—a bit like how humans entertain several ideas before deciding. The model is already being integrated into Google’s Workspace tools and coding products, with broader rollout planned for later this year.

The team behind Deep Think is aiming to get closer to “deliberative reasoning,” a goal that’s been long sought in the AI community. Oriol Vinyals, VP of research at DeepMind, said the approach lets the model “explore many possible futures before committing to a final answer.” This could make AI more useful in open-ended tasks like debugging code, planning projects, or writing essays—where there isn’t a single right answer. According to TechCrunch, it’s already showing signs of improved performance in benchmarks like Big-Bench Hard and HumanEval.

One striking implication: if AI can simulate creative ideation instead of reacting with fixed patterns, it could become less of a tool and more of a thinking partner. For knowledge workers, the future isn’t about automating the answer, but collaborating on the question. That changes how we think about productivity—not as output, but as cognitive exploration.

Takeaways at a Glance:

• Gemini Deep Think introduces multi-path inference: it runs several hypotheses in parallel

• Aims to simulate "deliberative reasoning" rather than reactive prediction

• Integrated into Google Workspace and coding tools already

• Benchmarks: Outperforms previous models on Big-Bench Hard and HumanEval

• Targets open-ended, ambiguous tasks requiring exploration—not just answers

• Possible implications for AGI-level reasoning and AI-human collaboration

What We Think About It:

• This model is a quiet but powerful step forward. By branching its thinking, it gets closer to how humans process ambiguity—great for planning, debugging, and ideation. Google’s approach could reshape how we design products around AI: not just faster chatbots, but smarter assistants.

What You Can Do Right Now:

• If you're using Google Workspace—like Docs, Sheets, or Gmail—keep an eye out for Gemini upgrades that feel more helpful, especially when drafting emails or brainstorming ideas. Try asking it open-ended questions or requesting multiple options—it’s designed to handle ambiguity better than before.

  In your everyday searches or AI use, experiment with asking “what are some different ways to think about this?” You might be surprised by how this version of Gemini explores ideas, not just answers. It’s not about speed anymore—it’s about insight.

Get more done with these AI tools

Scispace: The quickest way to read and understand scientific literature

• Searches and summarizes literature.

• Draft manuscripts with structured insights and citations.

• Checks plagiarism and peer reviews work.

• Generates publication-ready visuals (charts, infographics).

• Identifies patents and prior art.

• Streamlines grant searching and proposal drafting, and much more!

PromptPlex: The command center for all your AI prompts

• Organize Everything: Use nested folders and smart tags to build a library that makes sense to you.

• Work Faster: Create reusable templates with dynamic variables to stop editing and start generating.

• Discover Great Prompts: Access a curated library of high-quality, expertly-crafted templates.

Golpo: Turn docs into explainer videos

• Create narrated, whiteboard-style explainer videos from raw documents, outlines, or internal materials.

AI in the wild

That’s all for today, folks!

• Reach out to our audience by becoming a sponsor here.

• If you’re enjoying the newsletter, share with a friend by sending them this link: 👉 https://www.futureblueprint.xyz/subscribe

• Looking for past newsletters? You can find them all here.

• Working on a cool A.I. project that you would like us to write about? Reply to this email with details, we’d love to hear from you!