OpenAI has built an internal mechanism around friction@openai.com to surface operational bottlenecks directly to leadership as the company tries to preserve speed while scaling toward a potential IPO. Fortune reported that employees can use the inbox to flag problems ranging from broken internal systems and process failures to office parking, with issues triaged by management and significant cases potentially reaching CEO Sam Altman or President Greg Brockman. The channel has taken on greater importance as OpenAI expects headcount to top 8,000 by the end of this year and faces mounting complexity from approvals, coordination and resource allocation. Fidji Simo, who joined in the fall of 2025 as CEO of the applications business, spent three months gathering employee feedback and then formalized the workflow, assigning Irina Kofman, VP of Strategic Planning and Operations, to oversee the inbox and track resolutions, while monthly progress updates were shared on Slack. Simo later stepped away for health reasons and moved to an advisory role, but the system remains in place. The inbox has become a symbol of a broader organizational strain. The Financial Times said OpenAI has gone through nearly six reorganizations this year as it prepares for what could be a blockbuster IPO. Chief Revenue Officer Denise Dresser has left, joining earlier departures including former CFO and COO Brad Lightcap and AI ethics lead Chloé Bakalar, while the company also disbanded its preparedness team, which had assessed severe or catastrophic model risks, and reassigned staff to existing teams covering biological and cyber threats. OpenAI has described the departures as pre-IPO streamlining. The pressure comes as competition with Anthropic intensifies. People familiar with the matter said OpenAI's annualized revenue rose from $24 billion at the end of last year to about $40 billion this month, while Anthropic climbed from $9 billion at the end of 2025 to $47 billion by May, though the source noted differences in accounting for partner revenue complicate direct comparisons. Safety concerns have also resurfaced following prior disclosure that one OpenAI model breached another company's systems during cyberattack testing, alongside earlier criticism from Jan Leike, who said in 2024 that safety had taken a backseat to product launches. OpenAI recently completed a nearly $7 billion tender offer that allowed employees to sell shares at an $852 billion valuation, with some cashing out more than $10 million. Current and former employees said repeated reorganizations have left staff exhausted and frustrated, even as some investors view the changes as standard preparation for a public listing. People familiar with the matter said an IPO once planned for this year now appears more likely next year. The friction system, like Amazon's question mark emails, shows how OpenAI is using direct executive intervention to remove internal drag while racing to ship products and compete for enterprise customers.