Guardrails aren't just safety — they're what the customer is paying for

11x’s Prabhav Jain inverts the usual view: “Guardrails aren’t just to prevent bad stuff from happening. That’s what your customers are paying you for.” Even inside one product, every use case needs its own — a regulated financial-services prospect demands different guarantees than a mid-market SaaS customer, and those guarantees determine how the agent is allowed to write, who it can contact, what data it can touch, what it can say on a call, and how every decision gets logged. So guardrails must be “built per use case, configured per customer, and audited continuously.”

Schmidt generalizes this to governance: the vertical company becomes the control plane where permissions, auditing, what-the-agent-is-allowed-to-do, and what-it-actually-did converge, and it absorbs the regulatory complexity (HIPAA, SEC/FINRA, FRCP and bar rules) the end buyer can’t take on alone — something a horizontal player “can’t credibly do without becoming a hundred different verticals at once.” This is the offensive case for SaaS survives as the governance and coordination layer — determinism still rules, it depends on Policy enforcement must run independently of model cooperation — hooks, not prompt instructions, and it surfaces Permissioned inference is harder than permissioned retrieval — enterprise context graphs need reasoning-level access control as the hard unsolved part — all rolling up into The system of work is the moat, not the model — the model is fungible underneath.