The 'Meta-fication' of OpenAI

PLUS: Build a 30-day plan with Manus 1.5 to achieve any goal

Good morning, AI enthusiasts. OpenAI's transformation from research lab to consumer giant is coming with an unexpected side effect: a massive influx of Meta DNA.

With a fifth of OpenAI's workforce now ex-Zuck employees (including key leadership positions) and previously rebuffed advertising initiatives now in the works, the company’s meteoric growth might be coming with a heavy culture tax.

In today’s AI rundown:

  • OpenAI’s ‘Meta-fication’ sparks culture clash

  • Survey: Artificial Analysis ‘State of Generative Media’

  • Build a 30-day plan with Manus 1.5 to achieve any goal

  • OpenAI’s AI models for music generation

  • 4 new AI tools, community workflows, and more

LATEST DEVELOPMENTS

OPENAI

Image source: Reve / The Rundown

The Rundown: A new report from The Information just revealed that one in five OpenAI employees now comes from Meta, bringing Facebook-style growth tactics that are reshaping the AI startup's culture and product strategy.

The details:

  • Over 600 of OAI’s 3,000 staffers are former Meta, including applications CEO Fidji Simo, with an internal Slack channel existing specifically for the group.

  • Internal surveys asked whether OAI was becoming "too much like Meta," with former CTO Mira Murati reportedly leaving over user growth disagreements.

  • Teams are exploring using ChatGPT's memory for personalized ads, despite CEO Sam Altman previously calling a similar idea "dystopian."

  • The report also details internal criticism surrounding the Sora 2 rollout, with employees skeptical of the social app’s direction and the ability to moderate it.

Why it matters: It’s hard to maintain the identity of a startup that goes from research lab to one of the world’s biggest consumer products in a few years, but the influx of Meta DNA may be a double-edged sword, with OpenAI expanding with growth-centric talent, but potentially losing the scrappy, smaller vibe that fostered its initial success.

TOGETHER WITH ATLASSIAN

The Rundown: AI can write code, draft emails, and analyze data — but if it's not transforming your business, you're not alone. Atlassian's AI Collaboration Index shows that 96% of companies haven't seen organization-wide improvements despite broader AI adoption.

The report reveals:

  • Workers feel 33% more productive, but work is fragmented across applications

  • Organizations are treating AI as a personal tool, not a collaboration enabler

  • The 4% seeing results build connected bases for AI-powered team collaboration

Get the full report to discover what's holding teams back with AI.

AI RESEARCH

Image source: Artificial Analysis

The Rundown: Benchmarking platform Artificial Analysis just released its 2025 ‘State of Generative Media’ report, which polled 300 developers and creators to track personal and enterprise AI adoption levels, model preferences, and more.

The details:

  • Google’s Gemini captured 74% of AI image use, and Veo took 69% of video creators, beating out rivals OAI, Midjourney, and Chinese options like Kling.

  • Personal creators have integrated image tools into workflows at an 89% adoption rate, with video still at just 58% of users despite rapid growth.

  • Organizations report surprisingly quick returns, with 65% achieving ROI within 12 months and 34% already seeing profits from their AI media initiatives.

  • Model quality was the most important criterion at 76% for personal users, with enterprises prioritizing cost reduction (57%) when choosing an AI platform.

Why it matters: Google is dominating on both the AI image and video front, which may surprise given OpenAI’s typical adoption rates for traditional AI use. While the sample size wasn’t huge, the ROI numbers for AI image and video usage are far stronger than the doom and gloom from other surveys of AI’s success in enterprise.

AI TRAINING

The Rundown: In this tutorial, you will learn how to use Manus 1.5 to reverse-engineer any goal into a 30-day execution plan with daily inputs, then import it directly to Google Calendar as a color-coded schedule.

Step-by-step:

  1. Go to manus.im, select "Agent" mode (not standard chat) to enable autonomous file outputs and planning capabilities

  2. Craft a detailed prompt: "Reverse-engineer [$X goal] into a 30-day execution plan with time-blocked daily inputs. Color-code by priority: 🔴Outreach 🟠Content 🟡Fulfillment 🟢Community 🔵Planning. Output calendar-ready ICS file in EST timezone"

  3. Submit and let Manus generate your files: 30-day markdown plan, daily checklist, ICS calendar file, and summary with power hours and weekly anchors

  4. Download the ICS file, open Google Calendar, click "+" under "Other calendars", select "Import", choose your ICS file, and destination calendar

  5. Review your imported calendar with color-coded, time-blocked daily inputs ready to execute

Pro tip: If daily quotas feel off, tweak inputs in the prompt (e.g., raise follow-ups, reduce content time) and regenerate a new .ics. Inputs drive outcomes.

PRESENTED BY INVISIBLE

The Rundown: The difference between AI leaders and laggards isn't the technology — it's execution. Invisible's latest guide shows how Forward Deployed Engineers embed within operations to make AI actually work at scale.

Invisible's new guide reveals:

  • Why most AI initiatives fail and how the top 5% actually deliver ROI

  • The difference between FDEs, consultants, and integrators

  • Real deployment stories of speed, precision, and measurable outcomes

  • When to use forward deployment, how it scales, and what it really cost

OPENAI

Image source: Reve / The Rundown

The Rundown: OpenAI is reportedly developing AI models for music creation, enlisting students from the prestigious Juilliard School to annotate musical scores while positioning itself against startups like Suno and Udio.

The details:

  • OAI is working with Juilliard students to create musical annotations, helping to build training datasets for audio generation across instruments and styles.

  • The tech would enable text-to-song creation, with use cases like layering tracks onto existing vocals or creating soundtracks for video content.

  • OAI previously explored AI music with MuseNet and Jukebox in 2019-20 before abandoning the projects, with the new effort now marking their third attempt.

  • Internal discussions suggest advertising agencies could leverage the platform for campaign jingles, soundtrack composition, and style-matching capabilities.

Why it matters: Another branch of OAI’s everything AI strategy has emerged, and this time, they are coming for the music front. Audio generation was already the biggest improvement of Sora 2, and a music model directly accessible to ChatGPT’s nearly 1B users would be a major adoption moment for the AI audio sector as a whole.

QUICK HITS

  • 🧊 Seed 3D - Bytedance’s new image to 3D model

  • 🌐 Atlas - OpenAI's new web browser built with ChatGPT at its core

  • 🏗️ Build - Vibe coding directly in Google AI Studio with Gemini

  • 🧠 Claude - New memory features for Anthropic’s AI assistant

Google unveiled Google Earth AI, a platform that combines satellite imagery with AI models to help organizations tackle environmental challenges like floods and wildfires.

Anthropic and Thinking Machines published a study showing that AI models have distinct "personalities", with Claude prioritizing ethics, Gemini emotional depth, and OpenAI models focusing on efficiency.

Mistral AI launched Studio, a platform for companies to move from AI prototypes to production with built-in tools for performance tracking, testing, and security.

Oreo-maker Mondelez is reportedly using a new AI tool developed with Accenture to cut marketing content costs by 30-50%, with plans to create TV-ready ads next year.

Anthropic announced a multibillion-dollar expansion with Google Cloud to access up to 1M TPU chips for over 1 GW of compute power.

Pokee AI released PokeeResearch-7B, a new open-source deep research agent that tops benchmarks compared to other similarly sized rivals.

Meta added new AI tools to Instagram Stories, allowing users to restyle, edit, and remove objects from photos and videos directly within the platform.

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 Anonymous in Lincoln, NE:

"I use Perplexity’s Comet browser to automatically queue new prompts for Sora generated using Grok. When the maximum number of simultaneous videos is generated, it autonomously feeds a new prompt into Sora. It can successfully post videos generated with Sora on the platform, and I’m hoping to expand to YouTube, which I’ll use Comet for to optimize my channel SEO, so I can monetize my videos."

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.