Agents Invade The Office; Vibing vs Coding; More Clankers ... It Figures
Google Bets Big On Workplace Agents
What happened: Google rolled out Gemini Enterprise, a single platform where non-technical employees can create and deploy AI agents that plug into company data. It bundles no-code builders, prebuilt assistants (research, coding, support), and an agent marketplace.
Pricing lands at $30/user/month (with a trimmed Business tier at $21). On the same day, Amazon pushed a similar bet with its Quick Suite concept, underscoring a platform race to become the “front door” for enterprise AI.
And of course we all know OpenAI chasing the platform play as well.
Why it matters: The fight is shifting from “who has the smartest model” to “who removes the most friction at work.” If your team can stand up an agent in minutes - connected to Slack, drive docs, and CRM - adoption follows. Expect consolidation: companies won’t juggle five agent builders; they’ll pick one that plays nicely with security, compliance, and identity.
🧩 Jargon Buster - Agent Marketplace: A catalog of ready-made AI workers (think app store) you can install, connect to data, and set loose on tasks like intake triage or research summaries - without writing code.
Read more: The Rundown
“Vibe Coding” Won’t Replace Engineers (Says Engineer & Warp CEO)
What happened: In a Dev Day sit-down, Warp founder Zach Lloyd told The Rundown that flashy “vibe coding” tools are closer to WordPress-style builders than pro engineering tools. Real enterprises run on messy, massive codebases - where AI must reason across millions of lines, repos, and tribal knowledge. Well, duh.
Why it matters: The near future looks like human + AI. Developers who can direct agents, verify outputs and stitch them into CI/CD will ship faster than those who don’t use AI - or those who rely on it blindly. AI is a multiplier, not a substitute.
🧩 Jargon Buster - Vibe Coding: Prompt-driven app creation that “feels” right in a demo but often struggles with scale, tests, and maintenance once code hits production.
Read more: The Rundown
🤖 Figure 03: Big Leap In Humanoids - Still Mostly Demos
What happened: Figure unveiled Figure 03: higher-frequency vision, sensitive fingertip sensors, wireless charging, and a new factory (BotQ) aiming at 100,000 units in four years.
Why it matters: It’s the clearest push toward “physical AI.” But the real milestone isn’t a launch video—it’s reliable, unassisted work in warehouses, retail, and homes. Until then, it’s promise on paper.
🧩 Jargon Buster - Embodied AI: Models connected to sensors and motors, letting intelligence act in the physical world (not just produce text or images).
Read more: AI Secret
🌐 Comet Turns Web Pages Into Actions
What happened: Perplexity Comet, the company's new AI browser, can summarize threads, pull key takeaways from long videos, and even draft posts directly from what you’re viewing - no copy-paste gymnastics.
Why it matters: We’re moving from “read → think → act” to “see → act”. Tools that collapse steps save creators and marketers hours. Just remember the guardrails: double-check citations and tone-check posts before publishing.
The Rundown University has a tutorial for ya.
🧩 Jargon Buster - AI Browser: A web browser with an embedded assistant that can read, transform, and act on page content (e.g., summarize, generate posts, trigger workflows).
Read more: The Rundown
📰 Survey: With News, People Use AI More, Trust It Less
What happened: The Reuters Institute reports weekly AI use has nearly doubled year over year across six countries. Information seeking (research/questions) now edges out pure content generation. ChatGPT remains the heavy hitter; Google and Microsoft expose millions to AI summaries inside search. Yet only 12% are comfortable with fully AI-produced news; 62% prefer human-written journalism.
Why it matters: Individuals like AI for themselves - but don’t want it as their newsroom. Publishers have to thread the needle: use AI to accelerate analysis and formatting, while keeping humans accountable for reporting and judgment.
🧩 Jargon Buster - AI Summary: A generated overview box at the top of results that blends sources into a quick answer, often reducing clicks to original outlets.
Read more: The Rundown
🎬 Sora’s Viral Surge Continues
What happened: OpenAI says the Sora iOS app passed 1 million downloads in five days, outpacing the early ChatGPT trajectory - even while invite-only.
Why it matters: AI video is now mainstream entertainment. The upside: zero-cost production for creators and studios. The kicker: a rights and deepfake minefield that platforms must police as aggressively as they scale.
🧩 Jargon Buster - Generative Video: Models like Sora that synthesize moving images + audio from text prompts or references - fast becoming a new creative medium.
Tools & Launches
- Zendesk AI Agents — Claims 80% autonomous resolution with copilot/voice handoffs. Strong if you’re already on Zendesk; pilot on low-risk queues first and measure deflection quality.
- QA.tech - A fleet of QA agents that generate coverage and pipe debug context into pull requests. Nice complement to unit tests; keep humans in the loop for flaky suites.
- Retool AI AppGen - Turn natural-language specs + your data into secure, production-ready internal apps. Great for ops dashboards; lock down auth and audits from day one.
- The Brief - An “AI agency in your browser” that manages the full marketing lifecycle - from insights to creatives to optimization. Impact hinges on data connectors and brand guardrails.
- ProhostAI - All-in-one vacation rental copilot: guest messaging, cleanings, and upsells. Watch booking-site API reliability and service-level KPIs.
- Dia - Now open to Mac users. A voice-first assistant with AI dialogue that feels natural. Competes with AI browsers; shines for listen-over-read workflows.
- 1Password Secure Agentic Autofill - Requires human approval before AI agents use stored credentials—closing a glaring agent security gap.
- Figma × Gemini - Figma integrates Google’s Gemini for faster generations across creative flows. Expect speed gains in asset ideation and iterative comps.
Today’s Sources: AI Secret • There’s An AI For That • The Rundown • Perplexity • Zendesk • QA.tech • Retool • The Brief • ProhostAI • 1Password • Figure.