This startup CEO spent $13,000 AI bill on OpenAI’s Codex in a single month

In the rapidly evolving landscape of artificial intelligence, the term "tokenmaxxing" has emerged, highlighting the staggering amounts companies are investing in AI technologies to enhance productivity. A recent report from a consulting firm unveiled a shocking $500 million expenditure within a mere 30 days on Anthropic’s Claude, raising alarms about the escalating costs associated with AI integration. Now, Dan Shipper, the CEO of Every, a company entrenched in AI-driven media, software, and consulting, has disclosed his own significant expenditure. During an interview with Business Insider, he revealed that he incurred approximately $13,000 in overage charges on OpenAI’s Codex, marking it as his most substantial AI-related bill to date. Shipper humorously noted the reaction of his COO, Brandon Gell, saying, "I got some side eye from our COO on that one." When reflecting on the year-over-year increase in AI costs, Shipper simply stated, "Way more. Way, way, way, way, way, way, way, more." He explained that the company heavily depends on OpenAI’s Codex for automating his email correspondence. The AI assists in managing his inbox, organizing his calendar, suggesting meeting times, and drafting replies, although it cannot send emails without prior approval. Despite the hefty expenses, Every considers these costs as standard business operations rather than supplementary expenses. Shipper emphasized that every employee has access to AI tools as part of their role, with early access to both OpenAI’s and Anthropic’s models. Full-time staff utilize a basic AI subscription priced at $20 per month, while technical team members are on higher-tier plans costing around $200 monthly. In cases where employees exceed their AI usage limits, the company covers any additional costs. Shipper likened these "token budgets" to typical employee expenses, similar to laptops or health insurance. He noted that Every has seamlessly integrated AI into its daily workflows, yielding significant benefits. Shipper remarked on the potential for AI tools to enable employees to operate in managerial capacities, stating, "Very few people actually get the opportunity to be managers, because managing humans is very expensive and risky. I think many, many more people are capable of that than we think." However, he also acknowledged the inherent limitations of AI, emphasizing the necessity of human oversight in its applications.