Bret Kerr 🛡️🧠🛜

I turn complex AI and enterprise technology into clear buyer narratives, executive-ready messaging, and multimedia systems that help technical product adoption.

[Bret Kerr 🛡️🧠🛜 @BretKerr]

Every freelancer, agency, and machine shop on earth has had the same laminated sign on the wall for fifty years:

GOOD. FAST. CHEAP. Pick two.

It was never a slogan. It was a law of physics wearing a customer-service smile. Good and fast? Won’t be cheap. Good and cheap? Clear your calendar. Fast and cheap? Hope you weren’t attached to “good.”

The agentic age was supposed to break this. Infinite leverage, marginal cost to zero, the whole pitch.

It did not break it. Anthropic just shipped it as a slider.

Look at the new @claudeai effort control. One axis. Faster on the left, Smarter on the right — good-vs-fast rendered as a single draggable dot. And the third corner of the triangle, cheap, isn’t missing. It’s in the fine print underneath: 2m 58s · 1.2k tokens. Cost is the leg you don’t get to grab. It’s just the bill for wherever you parked the dot.

Drag right toward Smarter and you spend time and tokens — you’ve traded away fast and cheap. Drag left toward Faster and you’ve traded away good. The iron triangle didn’t die. It got a haptic interface and a resting position called Max.

This is the actual shape of the agentic age, and it’s funnier than the marketing. We spent a decade automating the project manager whose entire job was to look a client in the eye and say “pick two.” Now the client picks two by dragging a slider, and the invoice writes itself in the background.

@AnthropicAI didn’t beat the oldest tradeoff in commerce. They did something better. They turned it into a UI affordance, put it one inch from the send button, and made you choose your corner of the triangle on every single prompt.

Good, fast, cheap. Still pick two.

Now with a satisfying drag gesture.

[Bret Kerr 🛡️🧠🛜 @BretKerr]

🚨 Confirmed: The @AnthropicAI @claudeai tier gamification @theo mentioned to @davis7 on @NerdSnipePod is now live in the iOS app too.

I’m absolutely loving Opus 4.8 — but I don’t just vibe with it.
I run it strategically as my Creative Director + Senior Dev PM, paired with Codex. The token burn velocity is very real.

If you’re building serious stuff, this combo is insane.
Latest @NerdSnipePod episode goes deep on Opus 4.8 (worth every minute):
https://t.co/hkhWKYqrN5

Recent build with Claude Opus 4.8 giving creative direction to Codex: https://t.co/fDySbjF9g2

Take a stateless quick 6 question quiz to see the shape of your career matched against a corpus of AI founders such as @bcherny @ilyasut @demishassabis

Who else is cooking with Opus 4.8 right now? 👇

[Bret Kerr 🛡️🧠🛜 @BretKerr]

Double token knockout on the $20 tiers today — hit limits on both Claude Code and Codex.
Indie devs are gonna be rotating through SOTA agents like maxed-out credit cards at this point 😂

[Bret Kerr 🛡️🧠🛜 @BretKerr]

@Cloudflare @grok build compatability with @claudeai Code CC @NerdSnipePod

[Bret Kerr 🛡️🧠🛜 @BretKerr]

First time properly using @grok Build (the new TUI/agent) to ship real changes.Guided it through adding mobile collapsible toggles + swipe carousel to the https://t.co/lzOu0qiF2K homepage.
It explored the repo, made a plan, created a reusable component, and applied clean edits while I steered.Really impressed with the workflow so far.(SuperGrok beta)

[Bret Kerr 🛡️🧠🛜 @BretKerr]

@Cloudflare Def a solid Plan B or C if you hit the Claude or Codex wall

[Bret Kerr 🛡️🧠🛜 @BretKerr]

🚨 Confirmed: The @AnthropicAI @claudeai tier gamification @theo mentioned to @davis7 on @NerdSnipePod is now live in the iOS app too.

I’m absolutely loving Opus 4.8 — but I don’t just vibe with it.
I run it strategically as my Creative Director + Senior Dev PM, paired with Codex. The token burn velocity is very real.

If you’re building serious stuff, this combo is insane.
Latest @NerdSnipePod episode goes deep on Opus 4.8 (worth every minute):
https://t.co/hkhWKYqrN5

Recent build with Claude Opus 4.8 giving creative direction to Codex: https://t.co/fDySbjF9g2

Take a stateless quick 6 question quiz to see the shape of your career matched against a corpus of AI founders such as @bcherny @ilyasut @demishassabis

Who else is cooking with Opus 4.8 right now? 👇

[Bret Kerr 🛡️🧠🛜 @BretKerr]

@AnatoliKopadze “Demis Hassabis spent 50 minutes at Stanford saying things most CEOs would never say publicly” 🙄🥴 it was his standard interview he has done hundreds of times TBH https://t.co/iUVLHLPyDj

[Bret Kerr 🛡️🧠🛜 @BretKerr]

The $20 AI Subscription is an Illusion 🤯

While platforms keep the baseline professional entry point anchored at $20/month, the underlying economics reveal a heavily stratified system functioning as a strategic loss-leader.

Key Insights from the Data:

The $20 Anchor: Baseline professional access (like @ChatGPTapp Plus and @claudeai Pro) remains fixed to build ecosystem lock-in, even as capability tiers expand far above it.

Price ≠ Cost: Analysts argue the flat rate behaves more like a "monthly exemption" rather than reflecting the massive infrastructure expense.

Extreme Stratification: Power developers and enterprise pioneers are now seeing capability tiers jump to $100 and even $200 per month to offset server loads.

Massive Contraction: Projected ChatGPT Plus users are expected to plummet by roughly 80%, dropping from 44M in 2025 to just 9M in 2026.

#AI #GenerativeAI #OpenAI #Anthropic #TechEconomics #SaaS

[Bret Kerr 🛡️🧠🛜 @BretKerr]

🤖🗣️🤖Having @claudeai Opus 4.8 High art direct and project manage @ChatGPTapp Codex 5.5 High

Saves on CC tokens and plays to their strengths

[Bret Kerr 🛡️🧠🛜 @BretKerr]

🫳🧩🛜Take a 6 question quiz to see what shape your career is: https://t.co/2pQVDfjSCS

And which founder type you match shapes with: @demishassabis @bcherny @ilyasut @ch402 @tfadell @steipete @EthanHe_42

https://t.co/iUVLHLPyDj

[Bret Kerr 🛡️🧠🛜 @BretKerr]

https://t.co/mOqNJuGatP

[Bret Kerr 🛡️🧠🛜 @BretKerr]

@NerdSnipePod

[Bret Kerr 🛡️🧠🛜 @BretKerr]

@NerdSnipePod

[Bret Kerr 🛡️🧠🛜 @BretKerr]

@NerdSnipePod

[Lukas Aichberger @aichberger]

We unlocked the working memory of LLMs 💥

Reasoning in Memory (RiM) replaces autoregressive "thinking out loud" with fixed memory blocks that form a task-specific workspace for latent reasoning.

The key idea is simple: reasoning should happen inside the LLM, not in its output!

[Bret Kerr 🛡️🧠🛜 @BretKerr]

https://t.co/mOqNJuGatP

[Bret Kerr 🛡️🧠🛜 @BretKerr]

When you run out of agentic tokens mid-build, the vibe coding doesn't stop—it just gets real.

My automated loop crashed, so I had to improvise the pipeline manually to ship the latest breakdown for https://t.co/FPi4e9ViT2:

The Assist: @grok Beta 4.3 walked me through the raw React and Next.js page edits locally.

The IDE: Swapped over to @antigravity IDE, manually updated the components, and saved the files.

The Terminal: Grok fed me the exact bash commands to manually commit and push to Git.

The Edge: @Cloudflare automatically picked up the main branch, compiled via @opennextjs/cloudflare, and deployed it globally.

The infographic below shows the exact modern full-stack architecture keeping the project alive. When the credit loop pauses, the fundamentals—clean Next.js App Router composition and a seamless Git-to-Cloudflare edge pipeline—are what save you.

Stop waiting for flawless autonomous loops. Ship anyway.

[Bret Kerr 🛡️🧠🛜 @BretKerr]

Ran out of credits on Codex, @claudeai Code, and @antigravity mid-deploy for https://t.co/8yDI5mAMRH 💀

@GeminiApp Flash 3.5 just walked me through manually injecting components, fixing a broken Next.js sitemap array, and debugging a nested JSX tag mismatch—all step-by-step via a raw terminal.

Prerendered, committed, and pushed live.

Turns out that 2003 structural syntax muscle memory still works. It’s not that hard when you actually understand the pipeline under the vibe. 🛠️💻 #buildinpublic #nextjs @bcherny

[Ethan He @EthanHe_42]

I've left xAI. It's been quite a journey.
I joined when xAI was about to build Grok Imagine from 0 to 1 - no data, no infra, no model.
Three months later, we shipped our multimodal video model, followed by reference-to-video and video extension.
I'm grateful for the opportunity to grow, learn, lead a team, and help push the field forward.
What I do next will be very different. More soon :)