- Everyday AI
- Posts
- Ep 760: AI Change Management That Works: 5 Moves The Top 5% Make (Start Here Series Vol 21)
Ep 760: AI Change Management That Works: 5 Moves The Top 5% Make (Start Here Series Vol 21)
Anthropic just signed a $100B deal with Amazon, OpenAI launching GPT-Image-2 in ChatGPT, and Apple’s AI power play with Tim Cook out, and more
Everyday AI
April 21, 2026
👉 Subscribe Here | 🗣 Hire Us To Speak | 🤝 Partner with Us | 🤖 Grow with GenAI
Outsmart The Future
Today in Everyday AI
8 minute read
🎙 Daily Podcast Episode: Most companies are buying AI but keeping the same workflows, so nothing changes. In Start Here Series Vol. 21, we break down the five moves that actually drive results. Give today’s show a watch/read/listen to learn more.
🕵️♂️ Fresh Finds: AI Studio and Gemini link up, OpenAI is testing a new ‘Hermes’ agent platform, and Trump reveals OpenAI partnership. Read on for Fresh Finds.
🗞 Byte Sized Daily AI News: Anthropic just signed a $100B deal with Amazon, OpenAI launching GPT-Image-2 in ChatGPT, and Apple’s AI power play with Tim Cook out, and more. Read on for Byte Sized News.
💪 Leverage AI: Employees are using AI to do work faster, but companies haven’t changed the jobs, so the gains never scale. Keep reading for that!
↩️ Don’t miss out: Miss our last newsletter? We covered: Google is forming a strike team to fix its AI coding models, OpenAI is committing over $20B to Cerebras servers, Google is exploring a new AI chip deal with Marvell, and more. Check it here!
Ep 760: AI Change Management That Works: 5 Moves The Top 5% Make (Start Here Series Vol 21)
You think proper AI implementation is a technical problem for your company to solve? 🤔
Wrong.
It’s actually about people.
Maybe it’s because of the fast-pace nature of AI, and the fact there’s literally dozens of new AI tech drops each week that promise to change how we work, but the defecto response to showing ROI on AI always defaults to the technical side.
Yet, studies show building an AI-native organization is WAY more about change management than anything else.
We break it down, and show you the 5 moves the Top 5% are making to get it right.
Also on the pod today:
• 70% AI value is people 🧑🤝🧑
• Why upskilling can backfire 🚫
• Tear up old SOPs ✂️
It’ll be worth your 35 minutes:
Listen on our site:
Click to listen
Subscribe and listen on your favorite podcast platform
Listen on:
Here’s our favorite AI finds from across the web:
New AI Tool Spotlight – Cosine is the coding agent built for maintainability, visibility, and control across every developer surface, PageOn.AI is an AI Presentation & Infographic Maker for Any Idea, Harker is a Free, private voice-to-text that works anywhere you type.
Trump AI — Trump booted Anthropic from the Pentagon for being "too radical," swapping in OpenAI instead. Anthropic might still get back in after recent White House talks.
New Codex Features — OpenAI’s Codex is getting customizable pixel-art avatars and screenshot-powered memory features, making coding feel a lot more personal.
Google AI Upgrades — Google AI Pro and Ultra subscribers just got higher usage limits in AI Studio, plus access to new models like Nano Banana Pro.
Qwen3.6-Max-Preview Released — Alibaba just dropped Qwen3.6-Max-Preview, showing off sharper coding skills and beefed-up instruction following.
OpenAI Hermes — OpenAI is testing a new "Hermes" agent platform inside ChatGPT, letting users create persistent AI teammates that run nonstop.
Codex — OpenAI is teaming up with big consulting firms to help more large companies use Codex in their operations.
Kimi K2.6 Released — Kimi K2.6 just launched, letting you run massive agent swarms and long-horizon coding sessions across languages.
New Zealand Engineers — Engineers in New Zealand can now instantly query underground data using AI, helping them build safer and faster.
AI Education — Google is rolling out new AI tools for education, bigger NotebookLM limits, and a way to easily transfer your Google Photos when you graduate.
Sharepoint AI Skills — SharePoint just dropped its new AI-powered Skills feature, letting users customize and share best practices.
1. Anthropic and Amazon Ink $100B Deal to Power Claude AI Expansion 💸
Anthropic and Amazon just signed a massive new agreement, committing $100 billion over ten years to ramp up capacity for Claude, Anthropic’s flagship AI model.
The partnership secures up to 5 gigawatts of computing power, with new Trainium2 and Trainium3 chips coming online this year and next, aiming to meet surging enterprise and consumer demand. Claude’s platform will soon be available directly within AWS, streamlining access for over 100,000 customers and keeping Anthropic at the forefront of the AI cloud wars.
2. OpenAI Unleashes GPT-Image-2 in ChatGPT 🎑
OpenAI is launching its next image generator today in an afternoon livestream.
Early testers are buzzing about realistic, readable newspaper designs and intricate 10x10 grids that old models simply couldn’t pull off. While some users still gripe about resolution limits, this drop clearly leapfrogs competitors in overall detail and complexity.
(More on this tomorrow after the official launch.)
3. OpenAI Rolls Out Chronicle Preview for Mac Users 🖥️
OpenAI has quietly launched Chronicle, a research preview feature for ChatGPT Pro subscribers using macOS, aiming to make AI assistants much more context-aware and less needy for repeated instructions.
Available now in the Codex app, Chronicle taps into your screen (with your permission) to build “memories” that help Codex understand what you’re working on, but it comes with privacy caveats and is not yet available in the EU, UK or Switzerland. The tool processes screen data locally and temporarily, though users need to watch out for unencrypted storage and increased risk of prompt injection from malicious content.
4. Claude Launches Live Artifacts for Real-Time Tracking ⏰
Anthropic’s Claude AI assistant just rolled out a major update, letting users build live dashboards and trackers that automatically refresh with current data inside Cowork.
Everything you create is saved to a new Live Artifacts tab with version history, so you can revisit and update your work anytime, from any session. This feature is now available on all paid plans, with the latest Claude app update ready for download.
5. Apple names hardware chief John Ternus as CEO as AI reshapes tech 📲
Apple announced Monday that Tim Cook will step down and John Ternus, the company’s hardware engineering chief, will take the helm, signaling an urgent pivot as AI rapidly changes how people use devices.
Ternus’s deep hardware expertise matters because Apple ties its chips, devices, and software tightly together, which could let the company control where AI features run and how efficient they are. The move raises immediate questions about Apple’s broader AI strategy and how it will turn new capabilities into revenue, even as the company readies products like a foldable iPhone and rumored AI-enabled wearables.
Most companies aren't losing at AI because they picked the wrong model.
They're losing because they're using AI to speed up old org charts, old workflows, and old management habits. That works for a minute. Then the ROI disappears, the politics show up, and the rollout starts quietly eating the company from the inside.
Yeah. That.
We broke this down on Everyday AI, and the takeaway is way bigger than tooling. The winners aren't the companies with the fanciest stack. They're the ones willing to rewire how work happens, what managers do, and what employees are actually being paid to produce.
That's the part a lot of leaders still don't want to touch.
And that's exactly where the upside lives. If your competitors are redesigning work while you're still debating licenses, they aren't just moving faster. They're learning faster, reallocating talent faster, and building muscle you can't catch with one big rollout later.
1. Fund the people side first 🔥
This episode keeps coming back to one uncomfortable truth. Most companies are treating AI like a procurement exercise when the real bottleneck is people, process, and behavior change.
That's why the Boston Consulting Group framing matters. If 70% of AI value comes from people and processes, 20% from data, and 10% from tools, then a lot of companies have the whole pyramid flipped upside down.
And that broken spend shows up fast.
The transcript points to 97% of executives feeling personal AI wins while only 29% see enterprise ROI. Your people are finding value. Leadership just hasn't built the systems to turn those wins into company results.
If your AI strategy still starts with seat counts and vendor comparisons, you're probably overfunding the easiest part and starving the part that actually changes outcomes.
Try This
Pull your AI spend into three buckets this week: tools, data and process work, and people change.
Then ask one blunt question. Which of these is actually supposed to change behavior across the company?
If the answer is still “the tool,” you found the problem.
2. Tear up your old workflows ⚡
A lot of leaders still treat AI like cloud or SaaS. Add it to the stack, train the team, and keep the workflow mostly intact.
Nope.
This transcript makes the bigger point clear. AI doesn't just improve the work. It changes the job inside the work, which means old SOPs usually aren't something to optimize. They're something to rebuild from scratch.
That's why the McKinsey point matters. Workflow redesign had the biggest EBIT impact among the attributes tested. Not model selection. Workflow redesign.
So no, you don't win by dropping ChatGPT into steps three and four of a clunky eight-step process. You win by asking whether the whole thing should now be two steps, with AI doing the grunt work and humans stepping in where judgment actually matters.
Try This
Pick one workflow your team secretly hates.
Then ask one question: if we built this today with AI already in the mix, would we do it this way at all?
If the answer is no, stop optimizing it. Rebuild it.
3. Rewrite the job before AI does 🚀
This is where the transcript gets sharper than most AI strategy talk. The real friction isn't just training. It's identity.
People built careers on judgment, expertise, and being the one who knew how to do the work manually. AI doesn't just challenge the process. It challenges who gets to matter inside that process.
That's why usage metrics are weak. License logins and seat counts tell you who touched the tool. They don't tell you whether the company actually changed how work gets done.
The better signal here is role redesign. Moderna, BBVA, and JPMorgan all point in the same direction: weekly enablement, leader-first adoption, reshaped roles, and behavior reinforced through management systems.
If employees are quietly using AI to automate huge chunks of their work while leadership still measures the old output, that isn't an employee problem. It's a leadership problem.
Try This
Rewrite one job description this week. Just one.
Map what AI now handles, map what the human now owns, and redefine the expected output around that split.
Then do it again next week.
Reply