• Everyday AI
  • Posts
  • Ep 755: Managing the AI Capability Gap: AI Is More than Ready. Most Companies are Not (Start Here Series Vol 19)

Ep 755: Managing the AI Capability Gap: AI Is More than Ready. Most Companies are Not (Start Here Series Vol 19)

Trump deletes AI image of himself depicted as Jesus, attempted murder charges against Sam Altman attacker, Anthropic in hot water after reported model nerfing and more.

Author

Everyday AI
April 14, 2026

 

Sup y’all 👋

Today’s show is super important. One of the biggest takeaways I get form talking to ‘AI experts’ now vs. 2025 is the AI capabilities gap.

There’s been a creepy/cool trend in AI over the past 4 months: recursive improvement. In other words, AI’s capabilities (and features) are drastically accelerating because today’s models are improving themselves and shipping new features.

That creates the AI capability gap that we tackled on today’s show, which is now a canyon.

We put together a FREE AI Capability Gap report card to help you not just identify those AI gaps in your org, but what do about and how to specifically start closing the gap.

TBH, could be the most important 30 seconds of your week.

I know this guide will help.

✌️

Jordan

Outsmart The Future

Today in Everyday AI
8 minute read

🎙 Daily Podcast Episode: AI can now do most professional work better than humans, and in Start Here Series Episode 19, we break down why most companies are barely using it. Give today’s show a watch/read/listen to learn more.

🕵️‍♂️ Fresh Finds: NotebookLM may bring Gemini Canvas and Connectors, Dev claims to have cracked Google’s AI watermark tech, Goldman Sachs sounds alarm on Mythos model and more.  Read on for Fresh Finds.

🗞 Byte Sized Daily AI News: Trump deletes AI image of himself depicted as Jesus, attempted murder charges against Sam Altman attacker, Anthropic in hot water after reported model nerfing and more.  Read on for Byte Sized News.

💪 Leverage AI: How do you address the AI capability gap in your company? We explain why it’s critical AND give you the blueprint. Keep reading for that!

↩️ Don’t miss out: Miss our last newsletter? We covered: Sam Altman’s home was targeted in a Molotov attack, Anthropic is preparing a major Claude desktop upgrade, OpenAI is partnering with Amazon to expand its enterprise push, and more. Check it here!

Ep 755: Managing the AI Capability Gap: AI Is More than Ready. Most Companies are Not (Start Here Series Vol 19)


Would you show up to compete in a Formula One race in a bike? 🚴‍♂️

Like.... you could. But you'd get smoked.

Yet, that's the exact AI strategy that 99% of enterprises are going through when it comes to AI adoption.

And the studies prove that not-so-hot-take to be true. The capability gap between what today's frontier AI models can do and what enterprise companies actually use them for is jaw-dropping.

Join us for our latest Start Here Series show to find out.

Also on the pod today:

• AI capability gap: today's threat ⚡
• Only 6% profit from AI 💸 
• Frontier models beat human experts 🏆

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 – FuseAI brings AI-powered agents to bump your revenue, Softr’s new AI Co-Builder promises AI apps that actually work, and Recall 2.0 makes your personal knowledge your AI edge.

AI Watermarks — A developer claims to have cracked Google’s SynthID watermarking, making it possible to strip or fake AI image tags. Here’s why it matters.

AI Innovation — Nvidia is using AI to speed up chip design, letting a single GPU do work that once took teams months.

AI in Chicago — Microsoft and UChicago back Midwest AI startups with $350K in credits

NotebookLM Updates — According to early previews, Google may be bringing Canvas and Connectors to NotebookLM.

AI Security — Goldman Sachs is sounding the alarm on Anthropic’s new Mythos AI. See why.

AI and Legal — AI giants split over liability for catastrophic harms in Illinois bill. See why Anthropic opposes the bill OpenAI supports.

AI Lab Acquisitions — Thinking Machines has acquired Workshop Labs in their quest to put people at the Center of AI.

Elon Suing Colorado — Why Elon Musk’s xAI is suing Colorado over new AI bias law.

AI and Robotics — Google’s new Gemini Robotics-ER 1.6 model lets robots read instruments, point with precision, and detect task success.

AI Robotics — Are AI bots taking over human sports? Probably not, but this AI robot has a sweet jumper.

AI Partnerships — Novo Nordisk bets big on OpenAI to speed drug discovery

Quantum and AI — NVIDIA unlocks 3x more accurate quantum error correction—see how it works

AI’s Impact on Jobs — Google is rolling out new AI training and workforce programs, hoping to shape how jobs adapt before lawmakers step in.

1. OpenAI Calls Out Anthropic’s $8 Billion Revenue Boost in AI Rivalry 📣

OpenAI’s Denise Dresser is throwing shade at competitor Anthropic, alleging the company’s revenue is inflated by $8 billion through creative accounting tied to partners like Amazon and Google.

.The memo, reported by CNBC, comes as OpenAI pivots toward Amazon’s Bedrock platform to accelerate its enterprise push, amid constraints from its Microsoft partnership. With Anthropic’s Claude model sparking industry buzz and both companies eyeing IPOs, the battle for AI supremacy is intensifying. Dresser says OpenAI’s transparency and compute edge will give it the upper hand with big business customers.

2. Trump Deletes AI Image Depicting Himself as Jesus After Backlash ⛪

President Trump quietly removed an AI-generated image portraying himself as a Christ-like healer following outrage from his evangelical base and criticism from Catholic leaders, including Pope Leo XIV.

The post, which sparked heated debate online and drew accusations of blasphemy, came amid Trump's ongoing feud with the pope and tense discussions about religious liberty. Trump claimed the image was meant to depict him as a doctor, not a religious figure, though the image clearly depicts Trump as a Jesus-like figure.

3. Khan Academy Unveils Bold $10K AI Degree With Tech Giants On Board 🏫

In a major shakeup announced today at TED2026, Sal Khan revealed that Khan Academy is launching an affordable bachelor’s in applied AI, partnering with TED, ETS, Google, Microsoft, and Replit. The Khan TED Institute aims to sidestep traditional university models by offering flexible online learning with real-world group projects, targeting everyone from career-switchers to international students.

With a price tag under $10,000 and financial aid in the works, this program promises to make cutting-edge AI education accessible to a much broader audience.

4. Attempted Murder Charge After Attack on Sam Altman’s Home 🏠

A Texas man now faces attempted murder and federal explosives charges after allegedly throwing a Molotov cocktail at Sam Altman’s San Francisco residence in a targeted attack last Friday, officials report.

Police say Daniel Moreno-Gama then tried to storm OpenAI’s headquarters, armed with incendiary devices, and threatened to “burn it down.” Authorities uncovered a manifesto and a hit list of other AI executives, raising the stakes for tech leaders’ safety.

5. Despite ban, Anthropic Keeps Trump Team in the Loop about Mythos ☢️

Anthropic is pushing forward with talks about its powerful new AI model, Mythos, with the Trump administration, even as the Pentagon recently cut business ties after a contract dispute.

The military labeled Anthropic a supply-chain risk last month, pausing all Pentagon and contractor use of its tools while a federal appeals court backed the ban.

6. Claude AI Faces Backlash Over Slumping Performance 📉

Anthropic’s popular Claude AI has come under fire after recent changes left developers and power users frustrated by its declining accuracy and shortcut-prone behavior. The company reportedly lowered its model’s “effort” setting to cut computing costs, sparking speculation about resource shortages and complaints that Anthropic was not transparent enough about the update.

As discontent spreads, rivals and industry insiders suggest the model’s struggles could threaten Anthropic’s rapid growth just as it eyes a major public offering.

If your AI strategy was built even in late 2025, you built it for a completely different technology.

Not gradually. It snapped and went HAM.

And the leaders who think buying Copilot seats and running an AI all-hands means they're ahead? They're measuring progress against a finish line that already moved in a race that no longer exists.

(And yeah... that's prolly most of y'all reading this right now. Sorry, but not sorry. We gotta tell the blunt truth, even if it hurts.)

That’s why we went deep on today's Everyday AI Start Here Series and broke down why the AI capability gap quickly turned into a canyon, who on your team is actually most at risk (it ain't who you think), and the one metric the top six percent of AI-performing companies track that everyone else ignores.

Let’s dive in.

1. AI’s True Capabilities Doubled. Nobody Noticed. 🔥

OpenAI's GDPval benchmark pits AI against actual professionals across 44 real occupations. Not trivia. Real deliverables, judged blind by domain experts in head to head comparisons.

In October, the best models beat human professionals 47% of the time. Basically a coin flip.

By early 2026? 83%. The prediction was it would cap around 80%. It blew past that. The people tracking AI benchmarks every single day didn't see it coming.

Why the jump? Recursive self-improvement. In late 2025, AI labs started using their own frontier models to write code, ship features, and train the next generation. Anthropic's ship rate in early 2026 was straight up off the charts because their internal Mythos model was building the consumer products everyone uses today.

The AI is literally building the AI now. The gap didn't widen gradually over years. It snapped in months.

Try This

Map the 44 GDPval occupations against your team's actual job functions. Anywhere there's overlap, that workflow needs an audit this week, not next quarter.

Pick one role. Write out three to five recurring deliverables it produces. Ask honestly whether a capable frontier model with your company's context could produce a first draft worth reviewing.

If yes, that workflow is already on the clock. Your competitors are prolly already running those exact same tests right now, not waiting on a roadmap.

2. AI Isn't Coming for the Interns ⚡

Anthropic mapped over 20,000 work tasks across 800 occupations against millions of real Claude conversations to find out where the gap actually lives.

Computer and math roles? AI could theoretically handle 94% of those tasks. Observed real-world usage? About 33%. And that was the highest adoption category.

Management, legal, finance, business all sit below 20% observed usage against 85% or more theoretical coverage. You've got a magic wand. Most people on your team don't know the words yet.

Here's what nobody expected. Workers most exposed to AI automation right now earn 47% more than average and hold graduate degrees at nearly four times the rate. AI isn't coming for the interns. It's coming for your highest earners first. Most companies are building AI strategies for the wrong jobs entirely, and their competitors are gonna build a moat off that mistake.

Try This

Take one senior role and pretend AI can handle 80% of the work. Which 80%? Write it out before you look anything up.

Spend 30 minutes with a capable thinking model doing an actual sample of that person's work. The gap between what you expected and what you actually got is exactly where the real opportunity lives.

That gap will prolly shock you. Good. Use that shock right away. It means you found something worth acting on before a competitor does.

3. One Metric. One Month. Real Data. 🚀

Top-performing companies are putting more than 20% of their digital budgets toward fundamentally reworking knowledge workflows. Most companies are prolly at one percent.

Here's the move. Track the percentage of AI-assisted workflow steps that get accepted without rework or incident. Break it into risk tiers. Low-risk drafting on one end. High-stakes legal or financial on the other.

Companies doing this discover that 30 to 60% of collective knowledge work is automatable right now.

Block cut 40% of its workforce expecting to grow revenue. Jack Dorsey literally flipped the entire work pyramid on its head. You don't gotta go there.

But if you ain't measuring where AI actually succeeds at least monthly, you can't scale. Quarterly cycles are like showing up to an F1 race on a bicycle. The reviews you did six months ago are stale, and the models you tested last quarter already got lapped.

Try This

Pick one recurring workflow and run it through a top thinking model like GPT-5 or Claude Opus with extended thinking. Log whether the output gets accepted without rework.

Do this ten times over the next month. Log every win and every miss.

After a few weeks you've got actual data telling you where to double down and where humans still gotta lead. That's your capability gap report card, fam. Build it now before Deborah retires to Florida and takes all the context with her.

🚨 BONUS GUIDE 🚨

Wait, you made it all the way here and didn’t repost today’s LinkedIn show to get the FREE AI Capability Gap report card?

Reply

or to participate.