AI Spend

AI Spend shows organization administrators what your organization spends on AI models each month and the estimated value of your agents. Use it to find expensive agents, compare cost with estimated time saved, investigate changes in usage, and decide which agents or budgets to review.

Only organization administrators can open AI Spend. Other members do not see the tab or its data.

Open AI Spend

In Kindo, select Command Center.
Select the AI Spend tab.
Choose a month, and narrow the view with the filters if you want.

If the tab does not appear, see Troubleshooting.

What AI Spend Counts

These rules apply to every amount on the dashboard:

Model usage only. Spend is the cost of model tokens, in US dollars, priced as described in What Counts as Spend.
UTC calendar months. Months and days start at 00:00 UTC. The current month includes the usage recorded so far, including today.
Live data. The dashboard reads recorded usage each time it loads data. It does not refresh on its own; to see new usage, return to the browser tab or reload the page.
Agent usage. An agent’s spend includes all usage recorded for the agent, except direct API requests.
Who the usage belongs to. The User filter and Spend by User use the member who made the request. An agent run started by a schedule, a webhook, or an integration event runs as the agent’s creator. The Department filter uses the department recorded with the usage, so moving a member to another department does not move spend already recorded. The Group filter uses the member’s current user groups.

Choose the Month and Filters

The month selector opens on This month. The list goes back to the first month your organization had spend. Hover over the selector to see the selected period in UTC; for This month, it also shows the day of the month and when the month resets.

Use Department, Group, User, and Agent to narrow the view. Each filter selects one item, and you can search the list. A list shows only the items with spend in the selected month under your other filters, so an agent or member with no spend that month does not appear. Department also lists every active department.

Status narrows the view to agents with a budget status: All, Over limit, or Warning. See Agent Status.

When a filter is set, the Clear filters button (X) appears. Select it to clear every filter and keep the selected month.

The page URL keeps your filters and any earlier month you select. Copy it to share the view with another administrator. A link to This month always opens the current month, so the view it shows changes when the month changes. Table sorting is not kept in the URL.

Summary Tiles

Tile	What it shows
Total Spend	All spend in the view, with its token count. Includes API Spend unless an Agent or Status filter is set.
API Spend	Spend from requests to the Kindo API with an API key. It is part of Total Spend.
Chat Spend	Spend from chats that do not use an agent. It is part of Total Spend.
Total Hours Saved	Estimated hours saved by successful agent runs. See Hours Saved and ROI.
Total ROI	Estimated value of agent runs minus agent spend, with the return as a multiplier, for example 2.5X overall, or 2.5X in selection when a filter is set.

Each tile has a daily sparkline for the selected month. API and chat usage have no agent and no agent status. When an Agent or Status filter is set, API Spend shows a dash and no sparkline, Chat Spend shows $0, and each tile says which filter excludes it.

Hours Saved and ROI

Hours saved and ROI are estimates based on settings your organization enters, not measured working time. Review these settings before you compare agents by ROI.

Setting	Where	Default
Default Hours Saved Per Run	Settings → Organization → Agent Runtime. Organization administrators can change it.	1 hour
Default Hourly Rate	Settings → Organization → Agent Runtime. Organization administrators can change it.	$60
HOURS SAVED PER RUN, HOURLY RATE (USD)	The agent’s AGENT CONFIGURATION panel in the agent builder. Anyone who can edit the agent can change them.	The organization defaults

Hours saved per run can be 0 to 40. The organization hourly rate can be more than $0 and up to $1,000; an agent’s hourly rate can be $0 to $1,000.

AI Spend calculates ROI for each agent, then adds the results:

Hours saved = successful runs × hours saved per run. Failed and stopped runs save no hours. A run counts on the day it started.
ROI = hours saved × hourly rate − agent spend. Agent spend includes the spend of every run, including failed runs.
Multiplier = ROI ÷ agent spend. When an agent has no spend, it has no multiplier and shows a dash.

ROI covers agents only. Chat spend and API spend are not part of it. With a Department, Group, or User filter, a run’s hours count in proportion to the share of its spend that matches the filter.

Find Agents to Review

Most Expensive Agents lists the 5 agents with the highest spend. Lowest ROI Agents lists the 5 agents with the lowest multiplier among agents with at least $25 of spend in the view. Select Agent, Spend, or ROI to sort the 5 agents shown; sorting does not bring in other agents. To see more agents, export a CSV or ask Kindo in a chat.

Column	What it shows
Agent	The agent’s name.
Spend	The agent’s spend in the view. A status icon appears when the agent has reached a warning or block amount.
Against limit	The agent’s spend counted toward its budget in the month (see Spend and Agent Total), as a bar against its warning and block amounts, or a dash when it has neither. Hover over it for the amounts, the number of runs, and the average spend per run.
Daily trend	Daily spend for the month against the normal range. Marks show the days spend reached the warning or block amount.
ROI	The agent’s ROI multiplier, for example 2.5X. Hover over it to see the ROI in dollars and the spend it is based on.

Spend and Agent Total

When a filter is set, Spend becomes Spend (in view) and an Agent total column appears. Spend (in view), Daily trend, and ROI follow your filters. Agent total, Against limit, and the agent’s status always use the whole month and ignore the filters.

Agent total is the spend that counts toward the agent’s budget, the amount that Against limit measures. Without filters, hover over Against limit to see it; an agent with no warning or block amount shows a dash there instead. For the current month, the Budgeting tab shows the same amount, though it updates only after each spend check. Agent total counts only the requests that agent budgets count, so it can be lower than the agent’s Spend. With filters, Spend (in view) is usually lower than Agent total.

Agent Status

The status toggle above the panels shows how many agents are in each status: All, Warning, and Over limit. Select Warning or Over limit to filter the dashboard to those agents; select it again to return to All.

Status	Meaning
Warning	The agent’s spend counted toward its budget in the month reached its warning amount, but not its block amount. An agent that reached both appears only under Over limit.
Over limit	The agent’s spend counted toward its budget in the month reached its block amount.

Status uses the agent’s budget, or the agent defaults when it has no budget of its own. Department budgets do not set agent status.
Status uses the budget amounts as they are set today, also for earlier months. If you change an agent’s amounts, the status shown for earlier months can change too.
Status resets at the start of each UTC month.
Over limit shows that the block amount was reached, whether or not budget enforcement blocked any requests. Blocking requires budget enforcement.
All counts every agent created by a member of your organization that is not deleted, whether or not it had spend in the selected month.
With an Agent filter, the counts include only that agent. With a Department, Group, or User filter, the counts include agents whose creator matches the filter. A note below the toggle says so. Spend in the panels follows the member who made each request, so the agents counted and the spend shown can differ.

To change a budget or lift a block, see Budgets and Resume After a Block.

Trends and Breakdowns

Spend over Time shows daily spend in the selected month, split into input, output, and API spend. Hover over a day to see its amounts and total.

The Spend by Department, Spend by Model, Spend by Group, and Spend by User panels each list the top 10 by spend, with a daily trend. When more than one entry remains, the rest are combined in All Others; when exactly one remains, it appears as an 11th row. Hover over a bar to see the amount and its share of the total.

A member can belong to several user groups, so their spend counts in each group. The Spend by Group shares are of Total Spend and can add up to more than 100%. Members in no group do not appear in Spend by Group.

The band behind each daily trend is the normal range: the average daily spend in the 30 days before the selected month, plus or minus one standard deviation, with the same filters. It shows how usual a day’s spend is. It is not a budget.

Export to CSV

To get the data behind the dashboard as a file, export it as CSV. The file lists every agent with spend in the view and every entry in each breakdown, so it is not limited to the top 5 or top 10, and it has no All Others row.

Choose the month and filters you want to export.
Select Export CSV.
Select Email me the CSV.

Kindo emails you when the file is ready, with the subject Your AI Spend export is ready. Select Download CSV File in the email to download it. The link expires in 24 hours; after that, export again. If the export fails, Kindo sends an email with the subject Your AI Spend export failed. Try the export again, and contact support if the problem continues.

The file has one section for each part of the dashboard:

Section	What it contains
Report	The organization, the month and its start and end dates, the filters, when the file was made (UTC), and who requested it.
Summary	Total, API, and chat spend, tokens, hours saved, ROI, and agent counts by status, with daily averages and the normal range.
Daily spend	One row for each UTC day, with total, input, output, and API spend.
Agents	One row for each agent, highest spend first, with its status, spend, spend counted toward its budget, warning and block amounts, ROI, and hours saved.
Spend by department, Spend by model, Spend by group, Spend by user	One row for each entry, with its spend, tokens, and share of the total. As on the dashboard, Spend by group shares are of Total Spend. Spend by user also has each member’s email address.

Amounts are in US dollars. The file shows the data at the time Kindo made it, so for the current month it can differ from the dashboard later. Your organization runs one export at a time, and an export can also wait while other organizations’ exports run. When your organization records audit events, each request appears as CSV Export.

Query AI Spend in a Chat

The dashboard shows the top 5 agents and the top 10 of each breakdown. To answer other questions, ask Kindo in a chat. Kindo uses the AI Spend Platform tools to read the same data the dashboard shows. For example, you can list every agent above an amount, compare several departments, or rank models by tokens.

The AI Spend tools are available to organization administrators. Use them in a Kindo chat with Platform turned on, which it is by default in chats you start yourself (see Tools). MCP clients connected through the MCP API can call the same tools.

For example:

For October, list every agent with more than $100 of AI spend, with its ROI and budget status, highest spend first.

Keep these limits in mind:

Each query covers one UTC calendar month. To compare months, ask for each month.
A query returns up to 10 agents, departments, or other groups at a time, or one row for each day. Kindo can request more pages for most questions.
Status filters use the same budget rules as the dashboard.
The tools only read data. They cannot change budgets, settings, or agents.

See Platform tools for the exact behavior of each AI Spend tool.

Troubleshooting

Problem	What to check
The AI Spend tab does not appear	Only organization administrators see it.
An agent, member, or group is missing from a filter	The filter lists only items with spend in the selected month under your other filters. Choose another month, or clear the other filters.
API Spend shows a dash, or Chat Spend shows $0	An Agent or Status filter is set. API and chat usage have no agent, so these filters exclude it.
Spend and Agent total differ	See Spend and Agent Total.
An agent’s spend differs from the Budgeting tab	The Budgeting tab shows the spend counted toward the budget, not Spend. See Spend and Agent Total.
Total Hours Saved or Total ROI is lower than expected	Only successful runs save hours. Check the hours saved and hourly rate settings.
An agent’s ROI shows a dash	The agent has no spend in the view, so it has no multiplier.
An agent is missing from Lowest ROI Agents	The table includes only agents with at least $25 of spend in the view, and shows only the 5 lowest. To see more, export a CSV or ask Kindo in a chat.
An earlier month shows a different status than before	Status uses the budget amounts as they are set today. See Agent Status.
An earlier month is missing from the month selector	The list starts at the first month your organization had spend.
The export email does not arrive	Check your spam folder. An export can wait for other exports to finish (see Export to CSV). If no email arrives, export again.