OpenAI takes on LinkedIn with jobs platform

PLUS: DeepSeek prepping ‘self-improving’ advanced AI agent

Good morning, AI enthusiasts. AI has been catching a lot of heat for reshaping the job market, and the company at the forefront of the revolution is launching a new program to help workers adapt to the chaos.

With a new Jobs Platform and certification programs, OpenAI is stepping onto Microsoft-owned LinkedIn’s turf, with a goal of upskilling 10M Americans in AI fluency in the process.

P.S. — Our next live workshop is today at 4 PM EST. Join and learn how to create templated workflows to scale your AI automation. RSVP here.

In today’s AI rundown:

  • OpenAI’s AI jobs platform, certification program

  • DeepSeek’s ‘self-improving’ AI agent

  • Transform photos into 3D-style visuals

  • Google’s EmbeddingGemma for on-device AI

  • 4 new AI tools, community workflows, and more

LATEST DEVELOPMENTS

OPENAI

Image source: Ideogram / The Rundown

The Rundown: OpenAI’s CEO of Applications, Fidji Simo, just announced the company’s plans to launch the OpenAI Jobs Platform, designed to connect businesses with AI-skilled workers, alongside a new certification program for AI fluency.

The details:

  • The platform will match employers with AI-savvy job candidates, with dedicated tracks for small businesses and local governments seeking talent.

  • OpenAI partnered with Walmart and other employers to develop certification programs that teach different levels of AI fluency directly within ChatGPT.

  • Simo said the goal is to certify 10M Americans in AI fluency by 2030, with the program expanding on its previously launched OpenAI Academy resources.

  • The initiative coincides with White House AI literacy efforts, with tech leaders meeting in Washington this week to discuss workforce development.

Why it matters: OpenAI is positioning itself as both a disruptor and a solution provider, creating AI tools that transform jobs while building infrastructure to retrain displaced workers. The move also pits OAI against (Microsoft-owned) LinkedIn in the talent marketplace, creating yet another front for the two icy partners to fight over.

TOGETHER WITH BOX

The Rundown: BoxWorks 2025 brings together the minds transforming how organizations handle unstructured data. Learn firsthand how AI and intelligent agents are turning document chaos into strategic assets, with next-level security built in from the ground up.

Experience BoxWorks for:

  • Live demonstrations of AI-powered data extraction and workflow automation

  • Master classes to develop your AI implementation skills hands-on

  • Insights from Fortune 500 CIOs on their AI transformation journeys

  • The latest innovations from Box's ecosystem of AI partners

Last chance to register for San Francisco or virtual access — save 50% on in-person passes.

DEEPSEEK

Image source: Midjourney / The Rundown

The Rundown: DeepSeek is working on a new AI with advanced agentic capabilities, including executing multi-step tasks autonomously and self-improving, according to Bloomberg — with the Chinese startup aiming for a release in Q4 of this year.

The details:

  • The new system will handle complex workflows with minimal user input and “learn and improve based on its prior actions.”

  • Founder Liang Wenfeng aims to deliver the agent by the end of the year, while the company’s R1 successor still awaits release after reported internal delays.

  • The launch would follow agentic trends from AI leaders, including releases like ChatGPT Agent, Anthropic's Claude for Chrome, and more.

  • DeepSeek has remained relatively quiet of late, despite Chinese rivals like Alibaba and Tencent pushing aggressive release schedules.

Why it matters: R1’s ‘DeepSeek moment’ shook up the AI model world less than a year ago, but the anticipation for the lab’s next major release has been a waiting game. With broad agentic capabilities still struggling to live up to the ‘year of the AI agent’ moniker, DeepSeek could have another sector-altering launch up its sleeve.

AI TRAINING

The Rundown: In this tutorial, you will learn how to use Google’s Nano Banana model to recreate any room or environment in isometric view, giving you a bird's-eye perspective that reveals hidden details and creates visuals for content/design mockups.

Step-by-step:

  1. Go to gemini.google.com, toggle on "Tools", and select "Create Images" (with the banana icon)

  2. Upload any room photo and prompt: "Recreate this image in isometric view" —suddenly see details that weren't visible before

  3. Refine elements: "Make the room bigger," "Add punk rock theme with minimalist chandelier" — Nano Banana edits without regenerating the image

  4. Swap environments: "Change cityscape window to ocean view" or "Add natural sunlight and a door to another room" — perfect for testing interior design ideas

  5. Push further with VEO: Upload your edited image and prompt "Make this room lively by adding two dogs running through" to create a video with sound effects

Pro tip: Nano Banana is great for both content creation and interior design mockups. It's excellent at editing elements while keeping the rest of the image consistent.

PRESENTED BY CLOUDERA

The Rundown: Join data and AI leaders at EVOLVE25 NYC on Sept. 25th to discover how companies are turning AI potential into real business value. Cloudera will showcase how top enterprises unify, manage, and leverage data to power intelligent transformation anywhere.

At EVOLVE25, you'll experience:

  • A deep dive into trends for data, analytics, and AI with industry analysts

  • Customer-led sessions on enabling secure data architectures and analytics

  • Hands-on labs on implementing secure AI from development to inference

  • Exclusive announcements from Cloudera’s leadership

GOOGLE

Image source: Google

The Rundown: Google DeepMind released EmbeddingGemma, a new addition to its open-source Gemma model family that is efficient enough to run on consumer devices, letting apps search and understand text in 100+ languages without internet.

The details:

  • The model works fast enough for real-time responses while consuming less memory than a photo app, making it practical for smartphones and laptops.

  • Google built it to power offline search across personal files, messages, and emails, keeping sensitive data on-device rather than sending it to the cloud.

  • Developers can adjust the model's precision based on needs, choosing between accuracy or faster speeds depending on the specific application.

  • The system already integrates with popular developer tools and runs directly in web browsers, enabling privacy-focused apps that function completely offline.

Why it matters: Google’s timing positions models like EmbeddingGemma as critical infrastructure for the coming wave of on-device AI agents and assistants, enabling a new class of privacy-preserving offline apps. Any on-device release from Google also now has extra interest given the tech giant’s potential Siri-powered ambitions.

QUICK HITS

  • ☄ Comet - Perplexity’s AI-first browser, now available to all students

  • ⚙️ Warp Code - Warp’s SOTA agent for code review, editing, and projects

  • 📓 NotebookLM - Google’s research tool, with new customization features

  • 🔎 Jan-v1-edge - Jan’s powerful new AI search model

Atlassian announced the acquisition of The Browser Company for $610M, with plans to expand its AI-driven Dia browser with enterprise-focused integrations and security.

Warner Bros. filed a new copyright lawsuit against Midjourney, alleging unauthorized use of its characters, like Superman and Batman, in AI-generated images and videos.

Microsoft unveiled new AI education commitments at the White House AI Education Task Force meeting, including free Copilot, educator grants, and LinkedIn AI courses.

Lovable rolled out Voice Mode, a new functionality powered by ElevenLabs’ speech-to-text model that allows users to code and build apps via voice commands.

AI search startup Exa raised $85M in a new Series B funding round at a $700M valuation.

xAI CFO Mike Liberatore left the startup, becoming the latest in a wave of departures that includes co-founder Igor Babuschkin and general counsel Robert Keele.

COMMUNITY

Every newsletter, we showcase how a reader is using AI to work smarter, save time, or make life easier.

Today’s workflow comes from reader Matthew in Windsor, CO:

"I used Google's NotebookLM and YouTube to perform internal business consulting. I pick one or more experts in a field, select their YouTube videos that discuss more overall strategy rather than how-tos and put that along with our business information (website, mission/vision statements, offering, and other brand brief-like material) all into NotebookLM and have the AI perform quick (and free) analysis of my business from the perspective of the selected expert(s). Really helped with refining how we present/phrase ideas and offering."

How do you use AI? Tell us here.

That's it for today!

Before you go we’d love to know what you thought of today's newsletter to help us improve The Rundown experience for you.

Login or Subscribe to participate in polls.

See you soon,

Rowan, Joey, Zach, Shubham, and Jennifer — the humans behind The Rundown

Reply

or to participate.