Whats up whats uppp!
You're getting this email because you subscribed to aidstation. If you're new here, welcome. I share what I'm using, what I'm learning, and what's happening in AI every week.
A couple of you took me up on my offer from last week and replied with what you were stuck on. We went back and forth, I shared how I'd approach it, and honestly those were some of the best conversations I've had through this newsletter. So the offer still stands.
If you're stuck on something, reply and I'll help you directly. If enough people ask about similar things, I'll turn it into a full guide.
Let's get into it.
Is your team actually using AI? Be honest.
Making your company AI-first isn't about buying a subscription to ChatGPT and telling your team to use it.
That's not enough.
It's about giving your team the actual tools, teaching them how to use them, and making sure AI is built into how they work every day.
Not optional, not "when they feel like it." Standard.
The difference between a company that's AI-first and one that isn't is simple.
In one, every person on the team builds automations, sets up agents for their specific work, and creates flows that run on their own.
In the other, people are still doing everything manually while the tools to make them 5x faster already exist.
That gap is where the layoffs are landing. Dorsey just cut almost half of Block, about 4,000 people. Reuters counted over 61,000 AI-linked layoffs globally since November.
At Kleva, we're not cutting anyone. We're replacing tasks, not people. Each person does way more because agents handle the repetitive stuff. But we also made sure everyone on the team is actually using AI. Not just the technical people. Everyone. That's the difference.
The companies that survive this wave aren't the ones that cut the fastest. They're the ones that made their people dangerous with AI before the cuts became necessary. There's still time to be one of those companies. But that window is closing fast.
If you're not sure how to make your company AI-first, or where to even start, reply. I'll personally help you figure out the approach, the tools, and the first steps. That's what this newsletter is for.
What I've been using
Claude Code took over my n8n
When I started using Claude Code, I noticed I was opening n8n less and less.
Editing a scenario in n8n, dragging nodes, configuring connections, debugging why a webhook wasn't firing, all of that took way longer than just describing what I needed to Claude Code and having it write a script.
I still had some scenarios running in n8n because the visibility was useful, seeing the logs, monitoring what's running, understanding the flow visually. Though I was already thinking about migrating those too.
Then last week I asked Claude Code to help me check something in one of those scenarios, and that's when I realized n8n had launched an API that lets you interact with workflows programmatically.
Build them, edit them, debug them, all through the API.
What really helped is building CLAUDE.md and memory files for your automation setup the same way you would for a code repo. What each workflow does, how they connect, which APIs they hit, naming conventions.
You give Claude Code the context once and every interaction after that is faster because it already knows the system.
I went back to Chrome
I love testing browsers. Arc, Dia, Atlas, Comet, I've tried them all. I was on Atlas for a while because I liked the concept and what it could become.
But it wasn't there yet. Too many bugs, things breaking when they shouldn't. Almost everything I do is browser-based, my entire workflow runs through it, and I can't afford my browser failing me.
I think AI-native browsers could be the future. But we're not in that future yet, and it doesn't feel like OpenAI is prioritizing Atlas the way they'd need to for it to catch up. So I'm back on boring, reliable Chrome.
Unplug OpenClaw from your Claude subscription
Quick heads up if you're using OpenClaw with your Claude account. Anthropic is banning accounts now. Mine got banned today.
If you're running OpenClaw, unplug it from your Claude subscription and switch to the API directly, or plug in your ChatGPT account instead. Don't learn this the hard way like I did.
What caught my attention this week
Anthropic got banned. Claude hit #1 the next day.
The Pentagon gave Anthropic a deadline: remove the safety restrictions on Claude or lose your federal contract. They wanted Claude available for "all lawful purposes," including mass surveillance of US citizens and fully autonomous weapons.
Dario Amodei (Anthropic’s CEO)said no. His exact words: "Threats do not change our position. We cannot in good conscience accede to their request."
Trump called Anthropic "out-of-control," and signed an order banning all federal agencies from using Claude. Six months to phase everything out.
The same day, OpenAI signed a $200 million deal with the Pentagon. Sam Altman said his company has similar restrictions against mass surveillance and autonomous weapons. But he took the deal anyway.
Then something happened that nobody expected.
A "QuitGPT" movement exploded online. Reddit, X, everywhere. Over 1.5 million people committed to switching from ChatGPT to Claude. Guides for deleting ChatGPT accounts and migrating to Claude went viral.
Within 24 hours, Claude went from outside the top 100 to the #1 downloaded app in the US and Canada, passing ChatGPT and Gemini. Free users are up 60% since the start of the year. Daily sign-ups quadrupled.
A federal ban was supposed to kill them. It turned into the best marketing campaign of the year, and Anthropic didn't spend a dollar on it.
Whatever you think about AI and defense, the lesson here is simple. Standing for something, even when it costs you a contract, can be worth more than the contract itself.
Quick hits
OpenAI raised $110B - Biggest private round in history. $730B valuation. Amazon put in $50B, but $35B of that only kicks in if OpenAI goes public or hits an "AGI milestone," whatever that means. The definition was redacted from the filings. Sam Altman says AGI is "a near term thing now." The companies building AGI started writing it into contracts, pay attention. —
Claude Code Remote Control - Anthropic released a feature that lets you control Claude Code sessions from your phone. Start a task on your laptop, scan a QR code, keep working from anywhere. Your full environment stays connected. Available for Max subscribers. — https://code.claude.com/docs/en/remote-control
Nano Banana 2 - Google dropped a new image generation model. Pro level quality at Flash speed. If you're using AI for images, this is the new baseline. — https://gemini.google/us/overview/image-generation/?hl=en-US
That's it for this week.
If you want help figuring out how to make your team AI-first, or you're stuck on a specific workflow, reply. I'll help you figure it out.
-Ed
Did you enjoy this newsletter? Check out what we're building at Kleva.
And if someone forwarded this to you, you can subscribe here: