AI budget & cost caps

The AI Budget page shows how much AI your store is consuming this month - total tokens, estimated cost, and request volume, with a breakdown by capability, provider, and model - and lets you set a monthly cost cap on AI model spend. Find it under Settings → Data → AI Budget, or go to /app/<storeId>/settings/ai-budget. The header reads AI budget with the kicker “AI cost cap.”

Spend cap vs. plan

The cost cap here limits AI model spend; your plan and invoices are managed separately from Billing (the spend table links straight to it with a “Manage plan” action).

The three headline metrics

Across the top, three cards summarize the current calendar month:

Tokens this month - the total input + output tokens consumed across every AI feature for this store.
Estimated cost - the dollar value of those tokens, shown to the cent (e.g. $12.40). It is an estimate derived from per-model pricing, not an invoice.
AI requests - how many calls were made to the models this month.

The monthly cost cap

Below the headline metrics, the Monthly cost cap card shows a status badge - No cap set, Within cap, Near cap, or Cap reached - and, when a cap is set, a progress bar of spend against it. Owners and admins (anyone with the settings capability) can edit two fields:

Monthly cap (USD) - a ceiling on AI model spend. 0 means no merchant cap. Choose a limit from your expected traffic and the usage shown here; there is no universal ratio between AI cost and subscription price. When spend hits the cap the AI stops billing further new provider calls for the rest of the UTC calendar month. Calls already in flight can finish; shopper-facing paths return their configured unavailable or handoff behavior.
Warn at (% of cap) - the near-cap warning threshold (default 80%).

Click Save cap to apply. The page requires the settings capability; teammates without it cannot use this settings page. Crossing the warning threshold or the cap also raises the AI budget warning and AI budget cap reached notifications - see Notifications.

A platform-set ceiling can apply in addition to your cap. You may lower your limit but cannot remove or exceed that ceiling. If Save cap reports a ceiling, use an amount at or below it. The effective lower limit still applies when a merchant cap is unset. This spend control is separate from the plan conversation allowance and is not a new billing rate.

Step-by-step: set a cap

Open Settings → Data → AI Budget in the dashboard.
In the Monthly cost cap card, enter a dollar amount in Monthly cap (USD) (or 0 for unlimited).
Adjust Warn at (% of cap) if you want the near-cap warning earlier or later than 80%.
Click Save cap and confirm the badge updates.

Spend by capability, provider, and model

Below the cards, a table breaks the same numbers down per capability, provider, and model, sorted by cost (most expensive first). Columns are:

Capability - the kind of work the model did, such as shopper_chat, merchant_copilot, embeddings, recovery_email, content_studio, or image_generation.
Provider - the routed provider kind, kept separate so an Azure deployment and an OpenAI model with the same id do not merge into one row.
Model - the concrete model id that handled the requests (e.g. an OpenAI, Anthropic, or Gemini model). Shows unknown if the model wasn’t recorded.
Requests - call count for that capability/provider/model row.
Tokens - tokens consumed by that pair.
Cost - estimated dollar cost for that pair.

This breakdown tells you where your AI spend actually goes. If one capability dominates, that’s the lever to pull - for example, a high chat cost might justify a tighter monthly cap or trimming unused conversion flows on the AI Setup page.

Empty until merchant AI runs

Before any AI usage is tracked, the table shows “No AI usage tracked in the current month.” Widget and channel chat, merchant Copilot, catalog work, embeddings, recovery, content generation, and social-image generation, when enabled, can all populate the table. New stores should expect this page to be blank on day one.

How costs are calculated

Zubby records usage for centrally routed merchant AI calls and multiplies by the provider’s per-token price for that model to produce the estimated cost. Image generation uses a per-image estimate because it is not billed as chat tokens. Numbers are aggregated daily and summed for the month, so the figures here are an internal estimate to help you budget - your actual bill is governed by your plan. See Billing for what you are charged.

Keeping costs down

Set a monthly cost cap so a traffic spike can’t run away with your AI spend - the assistant pauses billing model calls once the cap is hit.
Turn off AI conversion flows you don’t use (guided selling, smart discounts, browse abandonment) on the AI Setup page so they don’t generate extra calls.
Keep your catalog and knowledge base tidy so retrieval is efficient and the agent doesn’t burn tokens on noisy context.

Tune the AI behaviors that drive most of this spend under AI setup & behavior. Manage your plan and see invoices on Billing, and compare plan tiers on the pricing page.