You’ve probably noticed Codex resetting its usage limits every few days. The common consensus is simple: the product broke, OpenAI feels bad, and they refund your credits. It’s a nice gesture. It’s also completely wrong.
I tracked 36 public quota resets between April and August. Only 40% were actually tied to bugs. The other 60%? They happened exactly when user counts spiked, when new features dropped, or when a competitor was busy shooting itself in the foot. The product got more stable, but the resets multiplied. Why? Because the apology has become an acquisition strategy.
Look at what happened in late June. Anthropic started banning thousands of Claude accounts over abuse allegations. Developers were terrified their access would vanish overnight. It was a migration nightmare for Claude, and a golden window for Codex. Did Tibo (Codex’s public face) run ads saying “Come to us”? No. He just hit the reset button 12 times in July.
When your competitor trips, don’t push them—open your doors. Every reset was a silent signal to the market: “Come test our product, we have the credits to spare.” It worked flawlessly. Codex went from 5 million users in May to 25 million by August.
But the real genius isn’t just the timing. It’s the currency. Think about traditional growth plays. Tencent throws billions in cash red envelopes. Alibaba buys you milk tea. People grab the cash, drink the tea, and leave. A cash coupon buys attention. Free compute buys habit.
Codex’s free credits are a closed-loop subsidy. You can’t withdraw them to your bank account. To use them, you have to open the IDE, connect your project, and actually run the code. Every “apology” forces a real product trial. Tibo isn’t just giving away money; he’s buying your integration with compute costs.
And he turned it into a spectacle. Instead of a quiet backend notification, Tibo made a show of hitting a physical “reset” button. Users started calling him the “Cyber Godfather.” They screenshot his posts. They share his memes. He turned a support ticket into an event marketing channel.
If you’re a product manager, don’t just copy the free credits. Your product probably sucks, and giving away free trials will just let people experience how bad it is. What you need to copy is the operational leverage.
Most teams operate in silos. Customer support handles bugs. Product writes changelogs. Marketing buys ads. They never intersect. Tibo’s strategy uses one physical button to do all three jobs. A bug fix becomes a user recall. A feature launch becomes a hands-on trial. A competitor’s crisis becomes a migration funnel.
The smartest marketing doesn’t feel like marketing. It feels like a gift. Stop treating your operational mistakes as pure costs. Connect your support, product, and marketing. Turn every incident, every feature drop, and every competitor stumble into a reason to pull users back in. That’s how you win without a budget.
FAQ
Q: Isn't this just a gimmick that only works because OpenAI has infinite compute?
A: No. The compute is a catalyst, not the strategy. The strategy is timing. Throwing free credits at a buggy product just accelerates churn. The magic is using the credits as a lever to pull users back during competitor crises or feature drops.
Q: How do I apply this to my SaaS without a physical reset button?
A: Stop treating bugs as pure costs. When you fix an outage, don't just email an apology—give them a reason to log back in and try a new feature you just shipped. Merge your support and product marketing into one action.
Q: Is Tibo just the next tech grifter?
A: He's a brilliant operator. While competitors burn millions on ads and cash red envelopes, he uses one operational lever to drive trial, retention, and migration. It’s not grifting; it’s just a vastly more efficient use of a growth budget.