"Billing either builds trust or it breaks trust." That's @Fynn Glover, CEO of Schematic, and I think he's spot on. Yesterday, we co-hosted an excellent Office Hours session, where several of you shared your own experiences (and challenges) of shifting to usage-based pricing. My biggest takeaway is that most teams still define Trust as Visibility (e.g., being able to see usage). Fynn views Trust as a combination of Visibility and Governance (e.g., customers being able to do something to control usage proactively). Fynn believes (and I agree) that in the age of agents, governance will become a requirement for any SaaS & AI tool to prevent excessive agent usage. He shared four things you can actually do to provide the full picture of Trust: 1. Turn the feature and spend cap ON by default. Fynn's favorite example of a company that does this well is Supabase. On paid tiers, they turn AI features on by default, but also turn a default spend cap on by default. When a user hits their covered limit, they gracefully degrade performance with clear alerting instead of a wall. This is a great example of protecting customers by handing over the controls rather than withholding the product. 2. Ship the controls before the meter. During Office Hours, one pricing leader discussed a recent AI product launch that received pushback from customers because governance requirements weren’t ready to go at launch. By giving admins permission controls 30 days before a credit-consuming feature goes GA, it allows them to plan usage accordingly, or turn the feature off to avoid surprises. 3. Talk openly about overages and throttling preferences. This one is psychological, not technical. Some customers want you to cut them off before a runaway bill. Others would rather eat the overage than break a workflow. The only wrong answer is not having asked. If you're not throttling, you both need to be aligned on what that means for the invoice.