AI Credits and Budget Guardrails

Updated June 12, 2026

CampaignSwift’s AI is built for agencies that can’t afford surprises, in behavior or in cost. This doc covers the cost side: how AI usage is metered, how budgets are enforced, and how every action stays auditable.

Credits: transparent AI cost

All AI usage (the AI Copilot and every agent) is shown as simple Credits. No opaque token math: one unit, visible usage.

Credits come with breakdowns along the two axes an agency actually manages:

Budget guardrails

You set a monthly limit on AI spend, with two enforcement levels:

  1. Warning threshold: when usage crosses the threshold you set, you’re warned. Spend continues; you decide whether to intervene.
  2. Hard cutoff (optional): if you enable it, AI usage stops at the cap. With the hard cutoff on, AI spend can never run away. The cap is a wall, not a suggestion.

Where usage stands against the budget isn’t something you have to go check, either: AI spend vs budget is one of the standing items in the Owner Digest, so the owner sees it whenever the daily brief has something worth sending.

The audit ledger

Every agent action is recorded in a complete ledger: what was done, who ran it, and the result. Combined with the propose → confirm → execute rule (every write action is proposed first and runs only after a human confirms), this gives you a full answer to “what did the AI do, on whose authority, and what happened?” for any client, any time. That matters for internal trust, and it matters when a client asks.

How this relates to your plan

Plan pricing ($29–$397/mo flat, not per seat) covers the platform tiers: clients, social accounts, seats, and storage. See the pricing page for current details. Every plan starts with a 30-day free trial, no credit card required, so you can watch real credit usage with your own clients before paying anything.

Practical setup

  1. Set a monthly credit limit that matches what you’re comfortable spending on AI across the agency.
  2. Set the warning threshold below it, so you hear about unusual usage while there’s still budget room.
  3. Decide on the hard cutoff: enable it if a guaranteed ceiling matters more than uninterrupted agent work at month-end.
  4. Review the per-client breakdown periodically. It tells you which engagements lean on AI most, which is useful pricing input.