Good morning, AI enthusiast.

A startup is pioneering a new frontier in mental health treatment, using AI to design psychedelic-inspired drugs that leave the hallucinogenic 'trip' behind.

By separating the therapeutic benefits from the overwhelming psychoactive experience, this AI-first approach could create a new class of precision treatments for mental health. But will this novel method be enough to navigate the tough regulatory hurdles that have challenged similar therapies in the past?

In today’s AI recap:

• How I Use Ahrefs MCP with ChatGPT & Claude (plus 4 SEO Agents)

• AI engineers psychedelics without the trip

• AI models pass the CFA exam

• The rising cost of AI ‘workslop’

• 8 trending AI Tools

AI's Psychedelic Frontier

The Recap: A Y Combinator-backed startup, Mindstate Design Labs, is using AI to engineer psychedelic-like drugs that deliver therapeutic benefits without the hallucinogenic "trip."

Unpacked:

• Mindstate built AI models that connect biochemical drug data to over 70,000 “trip reports” gathered from sources like clinical trials and online forums to map out specific psychoactive effects.

• Its first drug candidate, MSD-001, produced psychoactive effects like heightened emotions and enhanced imagination with no hallucinations in a 47-person Phase I trial, validating the AI platform's design.

• The company plans to uncouple its drugs from talk therapy, a strategy that could help it navigate regulatory hurdles like those that led the FDA to reject MDMA-assisted therapy last year.

Bottom line: This AI-first approach to drug design could create a new class of precision treatments for mental health. It opens the door to therapies that are potentially safer and more accessible for patients who want relief without an overwhelming psychedelic experience.

Marketing ideas for marketers who hate boring

The best marketing ideas come from marketers who live it.

That’s what this newsletter delivers.

The Marketing Millennials is a look inside what’s working right now for other marketers. No theory. No fluff. Just real insights and ideas you can actually use—from marketers who’ve been there, done that, and are sharing the playbook.

Every newsletter is written by Daniel Murray, a marketer obsessed with what goes into great marketing. Expect fresh takes, hot topics, and the kind of stuff you’ll want to steal for your next campaign.

Because marketing shouldn’t feel like guesswork. And you shouldn’t have to dig for the good stuff.

Subscribe Free

AI Tools of the Day

1. 🚀 Sonara – Automate your job search with an AI that finds and applies to relevant openings for you, landing interviews effortlessly.

2. 📈 Diagramming AI – Instantly generate complex UML and workflow diagrams from simple text descriptions to visualize your software architecture in seconds.

3. 📊 SheetAI.app – Supercharge Google Sheets with generative AI functions that can clean data, generate content, and answer questions using natural language.

4. 🧑‍💻 Greptile – Let an AI code reviewer with full codebase context analyze your pull requests, catching bugs and providing fix suggestions to help you merge faster.

5. 🎨 Recraft – Generate perfectly brand-aligned vector illustrations and 3D images by creating and locking in a consistent visual style across all your assets.

6. ⚔️ Friends & Fables – Embark on limitless D&D-style adventures powered by an AI Game Master that dynamically crafts stories and manages your campaigns.

7. 🤖 Octofy – Stop juggling AI subscriptions and let a smart agent automatically select the best model from ChatGPT, Claude, or Gemini for any given task.

8. 🎬 Wan 2.1 – Create high-resolution videos from text or images using a powerful, open-source framework that you can run and build upon yourself.

Explore the Best AI Tools Directory to find tools that will 10x your output 📈

AI Passes The CFA

The Recap: Top AI models can now pass all three levels of the notoriously difficult Chartered Financial Analyst (CFA) exam, according to new research from NYU Stern and GoodFin.

Unpacked:

• The major breakthrough is AI's ability to handle the Level III essay questions, a hurdle that stumped previous models, by using techniques like chain-of-thought prompting.

• Frontier models from Google (Gemini 2.5 Pro) and Anthropic (Claude Opus) were among the top performers, demonstrating advanced analytical reasoning for financial decision-making.

• While it takes a human several years and around 1,000 hours of studying, the latest AI models can successfully complete mock exams in just a matter of minutes.

Bottom line: This milestone shows AI is rapidly mastering specialized analytical reasoning far beyond simple pattern recognition. However, human expertise in understanding client context and non-verbal cues remains essential for now.

AI Training

How I Use Ahrefs MCP with ChatGPT & Claude (plus 4 SEO Agents!!)

The Recap: In this video, I walk through the step-by-step process for connecting Ahrefs MCP to Claude and ChatGPT. I also discuss 4 possible AI SEO Agents you could build with this new capability.

P.S. We also launched a free AI Automation Community for those looking to build and sell AI Automations — Come join us!

Alibaba's AI Arms Race

The Recap: Alibaba's shares surged after CEO Eddie Wu announced plans to increase the company's AI spending beyond an initial $53 billion three-year commitment. This move signals a major escalation in its push to dominate AI models and global cloud infrastructure.

Unpacked:

• CEO Eddie Wu projects the global AI industry will attract $4 trillion in investment over the next five years, framing Alibaba's spending as essential to keep pace.

• The investment supports a "full-stack" strategy, including the rollout of its new Qwen3-Max model and a global expansion of its cloud services with new data centers in Europe and South America.

• The spending increase follows strong performance, as Alibaba’s AI-related product revenue saw triple-digit growth and its cloud division's sales jumped 26% last quarter.

Bottom line: Alibaba is betting its future on becoming a foundational provider for the AI economy, moving beyond e-commerce to power global innovation. This aggressive spending reflects a belief that the long-term returns from controlling AI infrastructure will far outweigh the short-term costs.

Where AI Experts Share Their Best Work

Join our Free AI Automation Community

Join our FREE community AI Automation Mastery — where entrepreneurs, AI builders, and AI agency owners share templates, solve problems together, and learn from each other's wins (and mistakes).

What makes our community different:

• Real peer support from people building actual AI businesses

• Complete access to download our automation library of battle-tested n8n templates

• Collaborate and problem-solve with AI experts when you get stuck

Dive into our course materials, collaborate with experienced builders, and turn automation challenges into shared wins. Join here (completely free).

Beware of 'Workslop'

The Recap: A new term, “AI workslop,” is gaining traction to describe low-quality, AI-generated content that looks good on the surface but lacks substance. According to new research from BetterUp and the Stanford Social Media Lab, this trend is already costing companies millions in lost productivity as colleagues spend hours fixing it.

Unpacked:

• Each workslop incident takes an average of two hours to resolve, costing a 10,000-person company an estimated $9 million annually in lost productivity.

• Beyond the financial drain, receiving workslop negatively impacts team dynamics and trust, with nearly half of employees reporting they view the sender as less capable and trustworthy afterward.

• The problem largely stems from broad mandates to adopt AI without clear guidelines, encouraging workers to use the technology as a shortcut rather than a tool for collaboration.

Bottom line: The rush for AI-driven efficiency is inadvertently creating a new form of busywork that shifts critical thinking from the creator to the receiver. This highlights a growing need for clear AI usage policies that focus on thoughtful integration to unlock real value.

The Shortlist

OpenAI announced five new U.S. data center locations for its Stargate AI infrastructure project with partners Oracle and SoftBank, bringing its planned capacity to nearly 7 gigawatts.

NVIDIA released its Nemotron family of multimodal AI models, datasets, and techniques as an open-source resource for developers to build and customize agentic AI applications.

Anthropic integrated Claude Sonnet 4 into Apple’s Xcode 26, allowing developers to use the model directly within the IDE for tasks like debugging, refactoring, and generating documentation.

Google launched Mixboard, an experimental AI-powered app that allows users to create and edit visual mood boards using text prompts powered by its Nano Banana image model.

Signing off,

David, Lucas, Mitchell — The Recap editorial team