AI spend caps¶

Each plan includes a monthly AI spend cap in USD that limits how much analysis can run on Patcherly's platform AI keys. The cap is one of the included items on Pricing and shows up on Profile -> AI alongside your fix quota.

You see the cap because real AI provider tokens have a real dollar cost, and Patcherly absorbs that cost when you use platform keys. The cap exists so a runaway analysis loop or a very large prompt cannot quietly consume far more provider spend than your plan can support. Normal usage is nowhere near it.

When you see the warning¶

When your workspace has used 80% of the monthly AI spend cap, workspace admins receive an in-app notification and an email (if email notifications are enabled). The notification shows:

• How much you have used so far this period
• Your plan's cap
• When the period resets (Stripe billing date for paid plans, monthly anniversary for Personal)

Click the notification to open Profile -> AI, where the same numbers are shown with an Upgrade plan shortcut.

When you hit the cap¶

When you reach 100%, AI analysis on platform keys is paused for the rest of the period and the dashboard shows a Cap reached banner on affected errors. New manual analysis attempts return an error explaining the cap was hit.

To keep analysing errors immediately you have three options:

1. Upgrade to a higher plan from Profile -> Billing -> the cap refreshes against the new plan's ceiling on the next analysis.
2. Switch to BYOK (Bring Your Own Keys). Plans with Custom AI let you store your own OpenAI / Anthropic keys under Profile -> AI and turn on Use my own AI keys. While the toggle is on you pay your provider directly and your workspace's Patcherly cap does not apply. See Plans -> BYOK.
3. Wait for the period to reset. Paid plans reset on your Stripe billing date. Personal resets on a monthly anniversary from when the workspace started on Personal.

How spend is counted¶

The cap counts estimated AI provider spend based on the same token rates the Usage page uses (USD per 1M tokens per provider/model, refreshed on a weekly sync schedule). It is the platform's best estimate of provider cost; the authoritative numbers always come from your provider's invoices. See Usage analytics.

Spend that runs on your own API keys (BYOK with the toggle on) is not counted against the cap, but you can set your monthly cap directly through your own AI provider billing dashboard.

Platform-wide budget¶

Patcherly also keeps a monthly limit on the total AI spend across all free workspaces using platform keys, so one busy month doesn't crowd everyone else out. If that overall limit is reached:

• Free plans stop running new AI analyses on platform keys until the next calendar month. Upgrading to a paid plan resumes analysis right away.
• Paid plans keep working as normal. As long as you are still within your own monthly cap, your analyses keep running — paid workspaces are never paused because of free-plan traffic.
• BYOK workspaces are unaffected. Your analyses run on your own provider keys, so they don't count toward Patcherly's overall budget at all.

If you're on a free plan and you see a Cap reached banner that mentions the platform-wide budget rather than your own plan cap, the fastest way back to analysing errors is to upgrade or switch to your own AI keys under Profile -> AI.

Where to look¶

If you think the cap on your plan is too low for your team's normal usage, Open Support from the bottom of the dashboard sidebar — we may recommend the Pro plan / BYOK depending on your usage profile. Pay as you go / analysis credits plans are on the roadmap.