TL;DR
Sam Altman walked back his jobs apocalypse prediction the same week OpenAI is eyeing a $1 trillion IPO
Pancake is the AI infrastructure that runs your company ops across Slack, email, and iMessage while you focus on building
Claude Code now catches its own security vulnerabilities before they hit your PR
Figma shipped a native design agent that lives on the canvas and knows your design system
Google dropped a coding CLI and a no-code Android builder in the same week
CodeRabbit Agent builds a living memory of your stack so your team stops re-briefing the AI every session
Docusign API integrations can become harder to manage over time due to complex setup, enterprise-heavy processes, and rising costs.
BoldSign gives developers a simpler and more cost-effective way to add eSignatures to their applications.
With BoldSign, your team gets:
An AI-powered Docusign API to BoldSign API migration skill to make migration easier
Template and document migration support
SDKs for .NET, Node.js, Python, PHP, and Java
No complex go-live process
Transparent pricing with no hidden feature restrictions
Move from Docusign API to BoldSign API with less effort, lower cost, and a better developer experience.
THE AI JOBS APOCALYPSE HAS BEEN POSTPONED 🙄
For the past three years, Sam Altman has been one of the loudest voices warning that AI would wipe out white-collar jobs at scale. Entire job categories "totally, totally gone." Entry-level work disappearing overnight. He said it repeatedly, in interviews, on stages, in blog posts.
This week, as Sam typically does, walked that claim back.
Speaking at a Commonwealth Bank of Australia conference, Altman admitted he was "pretty wrong" on the social and economic impact of AI. White-collar displacement hasn't happened at the speed he predicted. "I'm delighted to be wrong about this," he said.
Ok cool…
The guy who built the thing that was supposed to end jobs is now saying jobs are fine. But look around. Uber burned through its entire 2026 Claude Code budget in four months. Meta just cut 8,000 people while pouring billions into AI infrastructure. Nvidia's VP of deep learning said AI might actually cost companies more than the humans they replaced. Yale's Budget Lab found no meaningful unemployment spike yet in high-AI-exposure jobs.
So which is it?
The disruption of AI in the workforce is real, but also uneven. It's not showing up in unemployment numbers yet. It's showing up in hiring freezes, in entry-level roles that never get backfilled, in teams of three doing what used to take ten.
Altman can afford to be delighted…most people don't have that luxury.
OUR TAKE
Altman went from "AI will eliminate entire job categories" to "I'm delighted to be wrong" in roughly twelve months. Smells like bullshit. Here you see a CEO managing optics ahead of a $1 trillion IPO that has to clear public market scrutiny.
The same week he walks this back: Uber's COO admits AI ROI isn't closing, Microsoft cancels internal Claude Code licenses, and Google AI Pro strips credits from paid subscribers.
The productivity gains didn't arrive at the pace the labs promised. AI layoffs mostly covered for cuts companies wanted to make anyway. And the largest enterprise customers stopped buying the original story.
None of this means AI won't eventually reshape how people work. But the loudest voices in the room are quietly tuning down their predictions…
QUICK HITS
DuckDuckGo installs up 30% after Google's AI search overhaul: Users are fleeing Google after it transformed search into a conversational AI engine with no real opt-out. DuckDuckGo, which lets you turn off every AI feature by default, saw iOS installs peak at nearly 70% week-over-week growth. Turns out people just want a list of links sometimes.
Claude Code now has a built-in security plugin that reviews its own code for vulnerabilities: It runs on three levels: file edits, model turns, and commits, catching issues before they become PR comments. Anthropic says internal rollout cut security-related PR comments by 30-40%. Install it from the plugin marketplace at /plugins.
Figma launched a native design agent that lives directly on the canvas: Prompt from any layer, run ideas in parallel, and let it handle bulk edits like renaming variables or swapping components across screens. It knows your design system, tokens, and components out of the box. Rolling out in beta now, free until general availability.
Google ships a coding CLI and a no-code Android app builder: Antigravity CLI brings the same agent harness to the terminal with full customization for your keybindings and workflow. Meanwhile Google AI Studio now lets anyone build native Android apps for free. Over 250,000 apps were created in the first week, most by people who had never written a line of Android code.
Your next campaign brief writes itself.
Most marketing teams spend Monday morning pulling numbers. Viktor spends it posting them. Cross-platform brief in #growth before the first standup. Spend anomalies flagged before they compound.
Your marketing team stops reporting and starts deciding.
THE BIGGEST PRODUCTIVITY KILLER
Stop re-briefing your AI every single session
⏱️ 15 min setup | 🔧 Slack + GitHub/GitLab + CodeRabbit Agent
Why build this? Every time your work moves from Slack to a PR to a terminal and back, your team loses context. Someone has to manually reassemble it. That senior engineer holding everything in their head? Single point of failure. Most AI tools make this worse, they're stateless. Smart, but they don't know your stack, your decisions, or what your team shipped last sprint. CodeRabbit Agent is built specifically to fix this. It lives in Slack and builds a living knowledge base from your actual team activity.
Steps:
Sign up for CodeRabbit Agent and connect your stack (GitHub or GitLab for code, Jira or Linear for tickets, Notion or Confluence for docs)
Optionally connect monitoring tools like Datadog, Sentry, or PostHog
Add the agent to your Slack workspace and invite it to relevant channels
Start a thread and ask something real: "What tickets are open in the login flow?" Watch it pull live context instead of guessing
Let it run. Memory builds automatically from your Slack threads and system activity over time
Expected outcome: After setup, your team stops re-briefing the agent every session. It knows what's open, what changed, and what your team decided last sprint. Runbooks get captured automatically. Tribal knowledge stops living in one person's head.
Full tutorial:
TOOL OF THE DAY
The AI cofounder that runs your company while you sleep
Most AI tools make you faster. Pancake makes you smaller (on purpose).
It's infrastructure for going from $1 to $1M without hiring. Memory, skills, schedules, agent coordination, all packaged so a solo founder or tiny team can punch way above their weight.
What's actually different here: it's not autopilot. Pancake surfaces decisions that need you and handles everything else across Slack, iMessage, and email. You set the direction. It runs the ops.
It also runs on open source agents (GitHub, forkable, inspectable) and passes tokens through at public API prices. No black box. No hidden markups.
Oh, and they run their own company on it. Outreach, content, engineering coordination, all Pancake. That's the kind of dogfooding that actually means something.
Solo founder? You get a full AI workforce. Small team? Everyone gets the same context and leverage. No one's working off a different version of reality.
