五度妙笔
API商城
数据库😺 Anthropic just LAPPED OpenAI
Sign Up
·
Advertise
Welcome, humans.
GPT Image 2 is like,
really good.
Here’s an example. Believe it or not, this is fake:
Here’s another one:
Now, what’s a practical use-case for this capability?
Interior design ideas!
Take a messy room, and envision what it could look like with different furniture and aesthetics.
So AI is
really
good at making images now.
So here’s an open question:
why can’t AI handle image files?
I actually think this is one of the biggest barriers holding back agents from producing web content at scale. Why can’t I give an agent an image file, say “insert this image in the document you create for me”, and then have it actually do it? We still have to do this manually, don’t we?
What’s up with that?
Here’s what happened in AI today:
😺 Anthropic just crossed $1T and passed OpenAI.
📰 Musk's lawyer told a federal jury OpenAI "stole a charity."
📰 Apple is shipping AI photo editing in iOS 27.
🍪 Proof lets Claude and Codex co-write your doc live.
📖 The AI buildout's costs are starting to show.
… and a
whole lot more that you can read about
here
.
P.S:
Want to reach 675,000 AI-hungry readers?
Click here to advertise with us.
P.P.S:
Love robots? We’re starting a new robotics newsletter!
Sign up early here
.
😺
Anthropic just LAPPED OpenAI.
Yesterday, Anthropic crossed a $1 trillion valuation and officially became the most valuable AI company on Earth.
The same Tuesday, OpenAI:
Missed its own revenue projections (per
WSJ
; Oracle and chip stocks fell)
Watched
Elon Musk's lawyer
tell a federal jury Sam Altman "stole a charity"
Pivoted onto AWS Bedrock
, one day after dropping Microsoft exclusivity
Got a piece in The Atlantic
literally titled "Anthropic's Little Brother."
Just normal Tuesday stuff.
Here's what
else
happened.
While OpenAI was getting sued, Anthropic launched
Claude for Creative Work
with native connectors for
Blender, Adobe Creative Cloud, Autodesk Fusion, Ableton, Splice, SketchUp, Resolume, and Canva.
Claude can now debug a 3D scene, batch-edit Photoshop assets, or pull stems out of Splice for you.
Adobe simultaneously shipped its own
Adobe for creativity Claude connector
, giving Claude live access to 50+ pro tools across Photoshop, Premiere, Firefly, and InDesign. Anthropic also
joined the Blender Development Fund
as Corporate Patron.
In one day, Anthropic became the AI you'd actually plug into the apps you already pay for.
Why this matters
: The race to "smartest model" is over. Both labs are locked in a feature-parity loop where Claude and GPT-5.5 trade benchmark wins every six weeks. The new race is "deepest workflow," and whichever lab gets buried inside the tools you already use wins.
The chat interface is saturated; workflow depth is the moat. And the deepest workflow integration possible is the computer itself. Once small models get good enough (and they will, in the next 18 months if not sooner), whoever's agent runs natively on your Mac, your iPhone, or the next device paradigm (glasses?) will control the surface where every other piece of software lives.
Our take:
This is why I now believe OpenAI & Anthropic will have to become Microsoft and Apple. Cloud tokens are an insanely expensive business right now.
An Nvidia exec said yesterday
the cost of AI compute has now passed the cost of employees at most companies. Neither lab can subsidize that forever.
The natural endgame? The one Apple and Microsoft already settled into.
Sell the hardware
.
OpenAI is reportedly building a phone
. Anthropic's creative-tool plays are workflow-deep but compute-light, the kind of thing that makes way more sense if Claude eventually lives on your Mac, not a data center. For this reason, we think Anthropic should eventually make computers too.
Today's $1T flip is the lagging indicator. The leading indicator? Who's best set up to sell you the device your local agent will run on.
FROM OUR PARTNERS
The era of building single-point AI tools is over.
Featuring insights from eight industry leaders, the new
2026 ‘Future of AI’
report reveals how and why startups must pivot to:
Orchestrate end-to-end agent workflows.
Deliver deeply personalized content.
Strategize around post-keyboard interfaces.
Establish grounded data benchmarking for your roadmap.
Harness your authentic human taste as your ultimate defensible moat.
Want to build a resilient, forward-looking foundation? You need this report.
Get report
🎓 AI Skill of the Day:
Make Claude Run Your Adobe Stack
The skill:
Use Anthropic's brand-new
Adobe for creativity connecto
r to make Claude orchestrate multi-app design workflows in plain English; no app-switching, no manual handoffs.
Why it matters:
Most creative workflows die in the gaps between apps (Photoshop → Premiere → InDesign → Express). The new connector gives Claude live access to 50+ pro Adobe tools across the full suite, so you describe the chain once and Claude does the steps.
The setup:
In
, enable the "Adobe for creativity" connector under Settings → Connectors. Then drop in a workflow prompt like the one below; Claude will surface drafts at each step and pass results to the next app.
You have access to the Adobe for creativity connector. Run this workflow for me, pausing for my approval between major steps.
INPUT: [Describe your asset; e.g., "a portrait headshot in my Lightroom library tagged 'speaker-headshot-2026'"]
WORKFLOW:
1. Photoshop: retouch the portrait (skin smoothing 30%, basic color correction, no aggressive edits).
2. Express: resize to 9:16 for Reels and 1:1 for Instagram feed.
3. Firefly: generate three thumbnail concepts pairing the headshot with "[YOUR TITLE]"; each in a different visual style (minimal, bold, editorial).
4. InDesign: assemble a single PDF one-pager with the retouched portrait, the three thumbnails, and a 100-word bio I'll provide.
Show each step's output before moving to the next. If you're not sure about a creative choice, ask me.
One favorite insight:
The connector lets Claude
reason about
your asset library. Try "find every product photo we shot in Q2 with a white background and resize them to a Shopify-ready 1500x1500." It works.
Want more tips like this? Check out our
AI Skill of the Day Digest for April
.
Total AI beginner?
Start here
(
goes with this video
).
Have a specific skill you want to learn?
Request it here.
Trending: Three popular Neuron podcast eps…
New episodes air
every week
on:
Spotify
|
Apple Podcasts
|
YouTube
 |
 |
 |
🍪 Treats to Try
Proof
is an agent-native realtime doc editor where you and AI agents (Claude, Codex, etc.) co-write the same file with separate identities and live cursors; no login needed —free to try.
Lovable's mobile app
lets you vibe-code full-stack web apps from your phone using natural language; it builds 20x faster than coding by hand —free to try.
SyncVibe
opens multiplayer coding rooms where you and your friends each plug in your own AI agent (Claude, Codex, or Gemini) and ship together over a shared invite code —free, open source.
Plurai's vibe-training platform
takes a description of your AI agent's intent and auto-generates edge-case datasets to train custom small models for evals and guardrails (43% lower failure rates, 8x lower costs vs GPT) —no pricing details.
Poolside open-sourced Laguna XS.2
, a 33B-parameter Mixture-of-Experts coding model (only 3B parameters activate per token, keeping it fast and cheap to run) that hits 68.2% on SWE-bench Verified; run it locally via
Ollama
(
ollama pull laguna-xs.2
) or call it through the
OpenRouter API
—free, open weights.
📰 Around the Horn
Anthropic began requiring photo ID and selfie verification
for some Claude users to enforce its usage policies; the move particularly impacts Chinese founders who'd been using VPNs.
Microsoft released World-R1
, an RL framework that injects 3D physical constraints into any text-to-video model for dramatically more geometrically consistent world simulations.
Freepik officially rebranded to Magnific
, positioning itself as the single pro creative platform; existing subscriptions continue unchanged.
The EU told Google to open up Android's AI
so other assistants can compete with Gemini's preferential placement; Google called it "unwarranted intervention."
Google and the Pentagon reportedly signed a classified deal
allowing the DoD to use Google's AI models for "any lawful government purpose" (which you might recall was the same legal language under dispute in the Anthropic DoD situation).
Want absolutely EVERYTHING that happened in AI this week?
Click here
!
FROM OUR PARTNERS
A reconciliation that takes 2 hours manually. Woodrow does it in 8 minutes.
Your team spends 2 hours on a single reconciliation — logging into systems, cross-checking line items, and emailing back and forth about discrepancies.
Woodrow
does the same job in 8 minutes, with every action logged and exceptions flagged for your review.
Learn more
📖
Midweek Wisdom:
The AI buildout is a bet that productivity gains will outrun the costs. Four reads on what's actually being wagered.
Kelsey Piper argues in The Argument
that AI's loudest critic Ed Zitron has "lost the plot" by ignoring the data-backed economics.
Zitron fires back on his Substack
that token costs run $8-$13 per $1 of revenue, data centers are overbuilt, labs are unprofitable, and there's still no productivity ROI.
"Why we need continual learning"
(a16z): Today's models are stuck in what the piece calls a "Memento and the Machine" loop; no persistent memory, no parametric updates from new experience. The argument: until models can actually learn from interactions, in-context tricks just paper over a deeper missing capability.
"Locked, stocked, and losing budget"
(The Register): AI vendor lock-in is biting back. Execs who thought they could swap foundation models in a week are discovering that the workflows, agents, fine-tunes, RAG layers, and compliance wrappers built around one vendor make migration prohibitively expensive.
New gas-powered data centers could emit more than entire nations
(Wired): A permit review of OpenAI, Meta, Microsoft, and xAI's planned gas-powered AI data centers projects 129+ million tons of greenhouse gases per year, about 1.9% of all US emissions and more than Norway or Switzerland's national totals.
A Cat’s Commentary
 |
|
P.P.S:
Love the newsletter, but only want to get it once per week? Don’t unsubscribe—
update your preferences here
.
