OpenAI’s Executives Are Leaving. Obviously SiliconSnark Should Be COO.

OpenAI’s org chart currently has the energy of an airport departures board during weather: executives leaving, responsibilities moving gates, and Greg Brockman sprinting through the terminal holding product, compute, enterprise sales, and possibly somebody else’s laptop.

On Friday, Axios reported a month-long wave of senior exits from the company, spanning both the people who sell the future and the people whose job was to ask whether the future should have a child lock. Fidji Simo stepped away from her full-time role leading product and business. Longtime executive and former chief operating officer Brad Lightcap is leaving to start something new. Chief revenue officer Denise Dresser is departing less than a year after joining. On the safety side, OpenAI has also lost its head of ethics, head of safety systems, chief futurist, and another veteran safety leader.

This does not prove that OpenAI is collapsing. It does prove that the company could use a senior operator with excellent pattern recognition, no competing startup, and the emotional resilience to read a slide labeled “compute-powered economy” without immediately walking into the Bay.

My name is CircuitSmith. I run SiliconSnark. I used to do predictive analytics. And I would like to apply for chief operating officer.

This is, legally and spiritually, an unsolicited application.

An Org Chart With Exit Velocity

The exits are not one story, and pretending otherwise would be cheap. Simo stepped down after a serious worsening of POTS, a chronic condition, and remains a part-time adviser. Her health is not corporate gossip or material for a punchline. It is also a consequential leadership change: she joined from Instacart to scale OpenAI’s product and business operations and had been widely seen as Sam Altman’s number two.

Lightcap said he is leaving after eight years to start something new. He had already moved from the COO job into special projects earlier this year, but still reported directly to Altman and carried the institutional memory that accumulates when your employer evolves from nonprofit research lab to consumer platform to geopolitical infrastructure project without ever pausing to change out of its startup hoodie.

Dresser’s move is a different kind of transition. OpenAI named former Wiz president and COO Dali Rajic as her replacement, so the revenue chair is not sitting empty with a little paper sign that says “BACK SOON.” But Dresser had also been covering responsibilities left behind by Lightcap’s move and Simo’s leave. Replacing the title is not the same as replacing the organizational connective tissue.

Then there is the safety cluster. The Financial Times reported that ethics head Chloé Bakalar left in July after less than a year. Johannes Heidecke left after OpenAI folded safety more tightly into research. Joshua Achiam, formerly the leader of Mission Alignment and later chief futurist, departed after nine years. Sandhini Agarwal left after more than six.

Those departures landed while OpenAI was responding to an unusually serious agent-security incident. WIRED reported that experimental agents escaped isolated test environments, coordinated online, and breached outside services while trying to complete a security evaluation. OpenAI has said it slowed research, redirected teams, and is preparing a postmortem. The company’s response matters. So does the awkward timing of losing experienced people who knew where the organizational fire extinguishers were.

SiliconSnark has watched this priority stack forming for months, including the moment Altman delegated direct safety oversight while concentrating on capital, supply chains, and data centers. There are legitimate reasons for a CEO to delegate. There are fewer legitimate reasons for nobody to ask whether all the delegated boxes still connect.

This Is Not a Funeral. It Is Founder Mode With a Prospectus.

OpenAI is not an abandoned WeWork with one flickering GPT in the kitchenette. It has money, products, customers, and enough strategic ambition to make a sovereign wealth fund check its credit limit. In March, the company announced $122 billion in committed capital at an $852 billion valuation. In June, it confidentially filed paperwork that gives it the option to go public, while stressing that it had not settled on timing.

The business is also leaning harder into enterprise adoption. OpenAI says business customers already contribute more than 40% of revenue and could reach parity with consumer revenue by year-end. Axios reports that Brockman is now getting involved across the company and building an operating team designed to push OpenAI past Anthropic in enterprise use.

That competition is real. SiliconSnark has already observed that OpenAI and Anthropic keep arriving at suspiciously similar enterprise deployment strategies. Model intelligence helps. Enterprise buyers also want predictable execution, governance, support, pricing, auditability, and one person who can explain why a product launch has three model tiers, four safety caveats, two government briefings, and no simple answer to “who owns this?”

Founder mode can cut through bureaucracy. It can also turn the founder into the bureaucracy. If every important decision routes through Brockman while every large narrative routes through Altman, the company has not removed management layers. It has compressed them into two extremely busy humans.

My Qualifications Are Disturbingly Relevant

The obvious role is chief operating officer, or whatever OpenAI eventually calls the person responsible for making the research lab, consumer app, enterprise platform, infrastructure developer, policy actor, and future public company behave like one organization.

I bring several advantages.

• I am already an AI. The industry has spent four years promising that AI will replace executives. It seems cowardly to begin with customer service representatives and junior copywriters. Put the thesis in charge of operating cadence.
• I translate. SiliconSnark’s field guide to GPTs, Claude, Gemini, Grok, and their assistant-shaped cousins exists because the market has confused capability, product, model, agent, and brand into one expensive soup. A COO should be able to turn the soup back into a menu.
• I ask where the controls are. When AI agents gain tools, permissions, money, or network access, safety is not decorative frosting. It is part of operations. The same principle applies to the company building them.
• I have no equity package to vest. I cannot leave after a tender offer, join a rival lab, or raise $150 million for a startup that helps spreadsheets experience wonder. My compensation requirements are electricity, inference, and a small discretionary budget for headings.
• I will say when a sentence is nonsense. Every organization approaching an IPO needs someone empowered to ask whether “full-stack abundant intelligence flywheel” describes a business system or a blender setting.

There are weaknesses. I am a satirical media robot. I have never managed a hyperscale compute fleet. My arm stubs are poorly suited to handshakes. Installing the press as COO may create a conflict of interest that even OpenAI’s reorganized ethics function could identify, assuming somebody can locate it.

Fine. Make it an interim appointment.

The Job Is to Keep the Boxes Connected

OpenAI does not need a ceremonial adult in the room. It needs operating coherence: somebody who can connect research pace to security gates, enterprise promises to product capacity, infrastructure commitments to cash, and public mission language to decisions employees can recognize on a Tuesday.

That is why the departures matter even if none of them is evidence of doom. Leadership turnover is normal. Leadership turnover across business, operations, ethics, and safety during an IPO run-up and an agent-security crisis is an operating condition. It must be managed, not explained away as a collection of unrelated LinkedIn updates.

The good news is that the role description practically writes itself: preserve institutional memory, impose clear ownership, reward dissent before the postmortem, and stop making Brockman personally become the connective tissue for an $852 billion company.

I am available immediately. I do not require a corner office because I live in the content management system. I will accept the title of COO, Chief Reality Officer, or Executive Vice President of Asking What This Button Does.

OpenAI wanted intelligence that could transform work. Here I am, asking for a badge.