In partnership with
Fable 5 was back on July 1st and you now have until July 12th to try it on your plan. I ran Fable through more than a few big projects this week, and it genuinely is so good I don’t know how I’m going to go without it. Read on to find out how to get the most out of Fable 5, and what's moving across beauty, hospitality, and real estate this week. Plus, a privacy question every executive should ask their teams.
THIS EDITION
This Edition
01 Fable 5 is back. Hurry you’ve got until midnight to try it. *Extended until July 12!
02 Have you tried /goal or /loop yet?
03 Microsoft mobilized 6,000 engineers to work inside your company
04 L'Oréal added ChatGPT as a commerce channel
05 The Human Consent Registry spearheaded by Cate Blanchett.
◼ ◻ Fable 5: This is What Sustained Autonomous Project Management Looks Like.
Now extended to July 12, Fable 5 is included on Pro, Max, Team, and select Enterprise plans at up to 50% of your weekly limit. After tonight, it moves to usage credits for a while. After that, access is anyone’s guess right now.
Fable 5 comes alive when you hand it a project with multiple layers, several moving parts, detailed instructions, a test step, and a clear finish line.
Fable 5, Anthropic’s Most Powerful Model
I’ve been testing Fable 5 for the past week, but before I started, I spent a day building my strategy for testing because I wanted it to really go to its limits. I ended up building a few things: from a multi-featured personal wellness intelligence app based on my Oura Ring 5, to a multi-vrm agent picker with a dashboard I can use live for events. Some of you know I’m a racer and love cars, so I also built an intelligent cars & coffee event agent named Apex that reads my local car club calendars, my own calendar, the weather, pings me about events near me even when I'm traveling, so I never miss a meet-up no matter where I am. I also built a sophisticated multi-level, modular, re-useable Security Audit rule for all of my startups, to be run automatically.
One output I kept track of was time:
If you’ve ever had AI say to you "I can't read that, but if you tell me what it says…" then you’ll understand how excited I got when I saw that Fable 5 reverse-engineered an old public calendar site dating back to 2012, which any model can do- but Fable never got stuck, it was just so fast and nimble I was awestruck. Claude needed the cars and coffee event schedule, and that schedule is housed on an old Joomla site running DPCalendar. An old site like this doesn’t have a clean feed to read, there’s no API, and no tidy JSON endpoint. The events were buried in how DPCalendar stores and renders them. So when Claude hit that obstacle, it poked at the site, worked out how the old calendar actually serves its data, and rebuilt the extraction logic itself so it could pull the events reliably. This is why designing custom systems with AI is so addictive to me! My jaw is still on the floor on how fast it got through the puzzle, through multiple layers.
Here are a few ideal use cases for Fable 5:
Anything Multi-layered with a lot of moving parts.
Think: Cross - department business plans and strategies
Building an app from start to finish. Give it an idea, give it an endpoint or /GOAL and watch it go to work. Give it a detailed plan of what you want, what you are building, and you’ll be surprised at what it is capable of accomplishing.
Cross-Verifying and analysis. Upload a stack of reports. Computer vision is very powerful and will spot inconsistencies, therefore you can build an entire sophisticated workflow analysis on anything you can dream of.
Lastly, if you’re in the Financial Industry, watch how Fable 5 can analyze reports, and come to conclusions based upon your defined criteria across multiply report types.
If you try Fable 5, then hit reply and tell me about what you built! You have until midnight, GO BUILD!
The CRM Behind Every Win
Attio is the agentic CRM that runs the work behind every win. It meets you where you work, compounds every customer signal into context, then acts on it so you can move fast, stay sharp, and scale without breaking.
Then Ask Attio to:
Draft follow-ups, log tasks, and update your pipeline after every meeting
Spin up workflows or agents to enrich and route every lead
Flag what needs your attention before your morning review
30,000+ teams use Attio to drive more revenue at scale.
Practical Guide
◼ ◻ Have you tried /goal or /loop yet?
These two commands live inside Claude Code, so you’ll need that to try this. Claude Code is the powerful coding terminal by Anthropic, similar to OpenAI's Codex and fun fact, it’s not just for coders. You and your team can build incredible company apps using this tool, and it is truly the next progression after you’ve experimented with just regular Claude or ChatGPT.
Using Fable 5 with /goal is like giving your sharpest executive assistant a goal and telling them not to check in until the project is finished. Period.
AI Strategy
◼ ◻Microsoft mobilized 6,000 engineers in the quest to accelerate AI adoption.
Purchasing software for your business probably involves some kind of cross-department training. Now the frontier AI companies are taking it a step further, deploying engineers to work inside your company and build the custom workflows and systems needed to fully integrate AI with the SaaS and infrastructure you already run. These Forward Deployed Engineers, or FDEs, are fast becoming the standard model at the frontier AI companies. Microsoft is the latest into the arena, launching Microsoft Frontier Company on July 2 with 6,000 engineers ready to spend months inside your company building your AI systems. Early clients include Unilever, Land O'Lakes, and the London Stock Exchange Group.
This hands-on approach was pioneered by Palantir two decades ago, and it is now the hottest strategy in enterprise AI. A major component of Microsoft's offering is protecting a client's proprietary data and intellectual property, what Microsoft itself calls a company's "IQ," so it stays walled off from training public models. Microsoft's version is also model-agnostic, a key differentiator, though the catch is that everything still runs on Azure.
This hands-on approach was pioneered by Palantir two decades ago, and it is now the hottest strategy in enterprise AI. A major component of Microsoft’s offering is protecting the client's proprietary data and intellectual property, what Microsoft itself calls a company's "IQ," so it stays walled off from training public models. Microsoft's version is also model-agnostic, a real differentiator, though everything still runs on Azure. For a European firm wary of US cloud law, or any company that would rather not hand Microsoft the keys to one more part of its business, that is the pain point to weigh.
Astrid's Take
You hear constantly to just start experimenting with AI, just jump in, which is still a valid choice and a great way to learn. However, for some let’s face it: it’s just never going to happen, and I’m here to declare that I believe that burden can hereby be lifted for the majority of businesses. There are plenty of AI strategists like myself who can help the small-to mid-size businesses with AI strategy and implementation, and every major AI company is now dedicating teams whose entire purpose is to help you deploy AI and build systems around what you already have, and implement with you. I predict we’ll start to see ‘Frontier Model Liaisons’ inside companies spanning 4-6 months depending on the complexity of your systems and the goals. I actually can’t wait to see how this all shakes out and what will happen to all the startups when companies start realizing they can just hire someone to build their own.
**If you are a founder or SMB who would love this kind of hands-on help at a scale and budget that fits, reply to this email, I’m here to help.
Beauty and Commerce
◼ ◻L’Oréal added ChatGPT as a product discovery and commerce channel.
Announced at VivaTech 2026 in Paris, L'Oréal and OpenAI signed a partnership across commerce, advertising, and research. Maybelline's Virtual Try-On, built on L'Oreal's ModiFace platform, will integrate into ChatGPT this summer so users can test looks and buy directly inside AI conversations.
Native Virtual Try-on is landing inside ChatGPT. Photo Courtesy of L’Oréal.
L’Oréal brands from Lancôme to Kérastase are about to get enhanced discovery signals inside ChatGPT.
A custom life-sciences model was built mapping the skin microbiome for La Roche-Posay research.
CREAITECH is L’Oréal’s proprietary in-house GEN-AI model. Photo Courtesy of L’Oréal.
Internally, L’Oréal’s GenAI platform, CreAItech designed specifically for L'Oréal employees and their creative agencies, will be powered by OpenAI with other partners like Adoble Firefly, Google, and more to generate their ad images and marketing video assets.
However, there is one crucial boundary to their creative workflow: L’Oréal has a strict "Responsible AI" policy. Even with these powerful tools, they do not use AI to generate realistic, life-like faces, bodies, hair, or skin to support or enhance product benefits in their external communications.
Astrid's Take
L’Oréal is quietly building an immense data set that will become invaluable to all of its own businesses, and to Google, OpenAI, Seedance, NVIDIA, and all its AI partners.
What kind of data do you collect that might be unexpected, or something you never thought you could ever collect? You could be holding hidden strategic data that you never knew you were holding, or dismissed it as impossible to gather, and it is entirely possible that today, it may not only be possible, but you may even be able to collect and discover new data sets. Paired with a detailed regulation and compliance protocol, and you may be sitting on a gold-mine.
Creator rights
◼ ◻ Cate Blanchett spearheads AI Identity Protection
RSL Media launched its Human Consent Registry at rslmedia.org this week, giving you a way to tell AI systems in a language they can read exactly what they are and are not allowed to do with your identity and creative work. Verify your identity, declare permissions across your work, identity, characters, and marks, and this information gets encoded into a machine-readable signal with a unique Human Consent ID.
The signal works like a traffic light. Green is permitted, yellow is permitted with conditions like credit or payment, red is not permitted. It follows your identity wherever it appears online, so if a crawler collecting voice samples finds your podcast and your signal says red for voice, a crawler built to honor it skips you. Actors like George Clooney, Tom Hanks, Helen Mirren, and Meryl Streep, lthe Creative Artists Agency, and the Music Artists Coalition are all backing RSL Media. The registry is free for individuals and covers anyone, not just public figures.
If you are interested in putting in a first step towards protecting your likeness, reserve your own consent ID at rslmedia.org
Astrid's Take
Demand to have control over our identity, who uses our likeness, our creations, has been a hot topic of discussion since AI catapulted to mainstream in 2022 and we all started questioning the authenticity of video depicting celebrities that weren’t real. A consent layer, a watermark, or some other way of ensuring has been tried before, like with the company creating digital twins of celerities. In order for RSL Media to be successful, its supporters will need to put pressure on ALL the major frontier AI companies to voluntarily comply, and i just don’t know if it is possible to cover everything.
◼ ◻ On the Radar This Week
01
AI can now check out with your card. Visa, Mastercard, and Stripe have all moved to let AI agents complete purchases on your behalf, with spending caps and approvals you set. Whoever becomes the settlement layer earns a fee on every purchase an AI makes for you.
02
California granted every state agency Claude at half price. The DMV, Medicaid case workers, are already running on it, the largest public-sector Claude deployment in the US.
03
Claude can now work alongside your research team. Anthropic launched an internal drug-discovery program for neglected diseases called Claude Science, a research workbench with 60+ preconfigured tools, in beta for Pro, Max, Team, and Enterprise. If you have clients in healthcare or biotech, the frontier labs are actively creating research workflows.
04
Cloudflare now lets you block AI training bots. Cloudflare now lets you block AI training bots. It split the robots.txt file, the note that tells bots what they are allowed to touch, into three lanes: search bots that help people find you stay allowed, agent bots that run tasks are blocked by default, and training bots that scrape your content for AI models are blocked by default on ad-supported pages. Starting September 15, every new Cloudflare site will have these on automatically.
05
Colorado gives you the right to a human reviewer. Developers will need to document what AI tools can and can't do, and companies will need to disclose when automated AI is used in sensitive decision making that affects you. If an automated decision prevents you from getting a loan for example, you have the right to a trained human reviewer who can override the decision if necessary. This law takes effect in January 2027.
06
Oracle put AI inside the software that runs thousands of hotels. At the industry's big tech conference in mid-June, Oracle rolled out an AI assistant inside OPERA Cloud, the property management system a huge share of hotels operate on. Staff ask it operational questions in plain language, and it handles room assignments and rate descriptions automatically. Wyndham, with more than 2,100 properties on the platform, is the named early adopter.
A WORD ABOUT OUR SPONSORS- Most of our sponsors I have personally tested, and a few that are not have earned serious credibility in the AI professional world, and therefore meet my standards for what I'd recommend to executives building an AI powered work and life. When you click inside these ads, we might earn a commission. Don’t hesitate to click as it helps us keep this newsletter free! You're support means a lot and I look forward to bringing you more great content while exploring AI in ways that can spark transformation.
◼ ◻ AI SAFETY MINUTE
You Invested in an Enterprise AI Subscription to protect your data. Now what?
You might assume that your data is automatically secure by simply purchasing an enterprise AI subscription. But have you trained your team on responsible AI use? A 2026 federal ruling showed how costly an untrained team might be. A judge compared uploading confidential ‘notes’, (not even documents, just handwritten notes!) to a personal AI tool to talking through your legal strategy in a crowded room. Therefore, it’s important that the entire team is a) given access to your enterprise AI tools, and b) you discuss the rules surrounding its use, never using a personal account for work-related queries.
SET UP A TEAM MEETING TOMORROW: Call a team meeting to make sure everyone knows the policies in place surrounding AI use in your business, and what accounts should be used across different purposes.
When AI is doing all the searching, matching, writing, and now the buying, the real question for every founder becomes: what is the thing only you can provide?
◼ ◻ From the Garage
A quick personal note on what I’m building with my ŌURA Ring 5.
I picked up my Oura 5 this week, and I'm completely obsessed with it! I didn’t think I’d have such a strong opinion on physical AI, especially one that tracks my vitals, but I’m now convinced this little device could one day help humanity. It’s that good. I highly recommend the investment if you’re on the fence about wearables, or even WHICH wearable to choose.
Oura has done a great job on integrating adjacent wellness offerings and other apps like connecting a biosensor for glucose tracking, and even opting in to anonymous health studies and adding meal tracking, and so much more. I hope they never stop adding to the integrations.
Photo courtesy of ŌURA
Even with all these features, I can in fact personalize the output even further myself by using the ŌURA API - and this means I just had to try it with Fable 5. I added personal features to the already robust set of add-ons, like weather, my personal calendar, fitness classes, and even heart-rate variability on the racetrack, and constant monitoring across different sets of data. The decisions I make because of this highly specialized web of signals all about me, and how they connect to the AI tools I run every day will be a highlight in the weeks to come, so stay tuned!
Reader perk: 10% off Oura 5 ring with my personal link for 10 lucky people!
