AI spend spikes
Overview
AI spend spikes automatically flag unusual increases in employee AI spending after each daily provider sync, show what drove them, and collect context from the people behind the usage—with no thresholds to configure.
Spend spikes are alerts, not budget limits or hard caps: responding to a spike records context and does not block AI usage or change a limit. For threshold-based alerts, see Set up AI Spend Limits.
What's included with AI spend spikes
- Automatic detection: Ramp compares each employee's daily AI spend with their typical spending and with their peers, then flags days that stand out.
- Spend spikes dashboard: Review spike counts, cost above typical spend, and response rates across your company. See Spend spikes admin dashboard.
- Spike details: Open any spike to see daily spend history, a breakdown by model, and the main spend driver when available. See Investigate a spend spike.
- Employee responses: Employees mark each spike as intended or unintended and can add a note, so you get context without following up one by one. See Respond to a spend spike.
- Individual alerts: Employees receive an email or Slack message when their own spending spikes. See Spend spike alerts for employees.
- Weekly summaries: AI Token Spend Admins receive a weekly recap of notable spikes and employee responses. See Spend spikes admin dashboard.
Spend spikes evaluate AI spend attributed to individual employees. For how usage is attributed, see Check how usage is attributed.
Who can use AI spend spikes
Spend spike notifications are available through an active AI Spend Management subscription. Access under existing or custom billing arrangements may differ. To open the dashboard, you need permission to view company-wide AI spend, and spend spikes must be available for your business. Your business must also have connected providers and synced data; see Set up AI Spend Management or the standalone setup guide.
| Who | What they get |
|---|---|
| AI Token Spend Admins | Weekly summaries. |
| Employees | Individual alerts about their own spikes, and the ability to respond to them. |
If you cannot open the Spend spikes tab, ask your Ramp Admin to review your access.
Spend spikes admin dashboard
Open the Spend spikes tab in AI Spend Management. The dashboard summarizes the selected time period:
- Total spend spikes: The number of spikes detected.
- Total cost impact: Spending above each employee's typical baseline, not the full amount spent on spike days.
- Alert response rate: The share of spikes that employees marked as intended or unintended.
Charts show cost impact or spike count over time, along with the top employees and departments by cost impact. Filter by time period, employee, department, spend driver, response status, or review status.
Select See all in Spend spikes this period to open the full table. Each row shows the date, employee, total daily spend, baseline multiple, driver, the employee's reason, and status. Spikes without a response after 7 days show No response.
AI Token Spend Admins also receive a weekly email or Slack summary of the most notable spikes from the past week, including employee responses. Select Review spikes in the summary to open this dashboard.
Investigate a spend spike
Select a spike in the Spend spikes table to open its details:
- Spending comparison: The amount spent, compared with the employee's typical daily spend or their department average, when available.
- Spend driver: The provider, product, or API key behind most of the spend, when available.
- Usage breakdown: A daily spending chart and spend by model.
- Employee response: Whether the employee marked the spike as intended or unintended, and their note, if submitted.
Use the response alongside the usage data to decide whether the increase needs follow-up.
Spend spike alerts for employees
When Ramp detects a spike in an employee's own AI spending, it sends that employee an alert by email or Slack, based on their notification preferences. The alert shows the spike date and amount, a comparison with typical spending, and the main spend driver when available. It includes an Explain your AI spend button so the employee can respond.
Alerts are sent after provider data syncs, not as each AI request happens.
Respond to a spend spike
Employees can respond only to their own spikes. Admins can see responses but cannot respond on someone's behalf.
- In the alert, select Explain your AI spend.
- Select Yes, intended or No, unintended.
- Optionally, add a note explaining what caused the increase.
- Select Mark as intended or Mark as unintended.
The response appears on the dashboard and in weekly summaries. If the response does not save, try again before closing the review.
Manage spend spike notifications
Individual alerts and weekly summaries have separate notification preferences, each with its own email and chat settings. To review them, see Manage your communication preferences.
To receive alerts in Slack, your business must first connect Ramp to Slack. Connecting Slack does not change anyone's notification preferences.
Frequently asked questions
Do we need to set anything up to detect spikes?
No. Spike detection is automatic; see Overview.
How are spend spikes different from AI Spend Limits?
Spend spikes flag spending that is unusual for an employee. AI Spend Limits send alerts when spending reaches an amount you configure. You can use both. See Set up AI Spend Limits.
Does marking a spike as intended approve the spend?
No. A response records the employee's explanation for reviewers. It does not approve spending, block AI usage, or create a spending limit.
Why didn't someone receive a notification for a spike on the dashboard?
Check that:
- Your business has an active AI Spend Management subscription or other access to spend spike notifications.
- The recipient's email or chat preference for that notification is turned on.
- For Slack, your business has connected Ramp to Slack.
- For weekly summaries, the recipient has the AI Token Spend Admin role.
If notifications still don't arrive, ask your Ramp Admin to review access, or contact Ramp Support.