A background agent did not do what you expected
An agent that did nothing is usually paused for credits, capped by its daily run limit, holding changes for review, or waiting on a trigger that never fired. The dashboard tells you which.
Open the agent and read two things before changing anything: the status in the header of its most recent run, and the Usage History on the Background Agents page. Between them they explain almost every case. See Monitor agent runs for how to read a run in detail.
Nothing ran at all
| What you see | What is happening | What to do |
|---|---|---|
| "No runs yet" on an agent you expected to have run | The trigger conditions have not been met, or the agent is paused. | Check the Schedule under "Agent setup", confirm the agent is active, and use the manual run to test it. |
| Every agent stopped and admins got an email | The workspace ran out of credits. A run that cannot be paid for records a failed attempt and deactivates the agent so it does not keep firing. | Top up under Settings > Billing, then resume each paused agent with its play button. |
"Daily limit reached" when you click Run agent now | The agent's daily run limit for today, counted in UTC, is used up. | Wait for the UTC day to roll over, or open the agent and raise or clear "Daily run limit" in Configure. |
| "Run not detected. The agent run did not appear." | The run was accepted but did not show in the list within 30 seconds. | Wait a moment and reload the dashboard. If it truly never starts, check the credit balance and whether the agent is paused. |
Clicking Run agent now asks for a record | Event-based and on-demand agents run against a single record. | Pick the record in the search dialog, or run the agent from Tables by selecting records and using the Run agent menu. |
| "Could not load runs" | The run history is temporarily unavailable, typically during a deploy. It does not mean the runs are gone. | Click Try again, then reload. Completed runs are still recorded. |
It ran but wrote nothing
A completed run alongside "No writes waiting for review" is not a failure. It means the agent looked and found nothing it was confident enough to change. Open the Run plan and the run's activity flow to see what it actually did.
Two settings commonly make an agent do less than expected:
- "Maximum records per run" caps the fan-out and defaults to 50. Any remaining allowance under the daily run limit trims it further. The Records tab's "Up next" list shows exactly which records the next run will take.
- "Run once per record" keeps an agent from reprocessing the same records forever. If it is off, a record stays eligible after every successful run. Turn it on in Configure and pick a cadence. The Records tab's "Already run" list shows the result.
Changes waiting for approval
An agent in proposal mode stages its CRM writes instead of applying them, so the work is done but nothing has landed in the CRM yet. Open the agent, review the proposed changes and approve them. If proposal mode is on and there is nothing there at all, either the agent has not run yet or it made no changes on the runs it did. Agent permissions covers what each mode is allowed to do.
Not enough credits
"Not enough credits to run" or "Not enough credits to resume", with the buttons greyed out, means the remaining balance is below what the agent needs to start. Top up at Settings > Billing and retry. A team credit budget can produce the same result on its own: when a team's monthly cap is reached, its members are blocked until the cap is raised or the month resets.
A run that will not finish
A run showing "Running" for a very long time has stalled without reaching a final status. Open it and click Stop process, or leave it: any run still active after 12 hours is displayed as Canceled automatically. Completed steps and their evidence stay readable either way.
Reading a failed run
Start at the run header: it carries the status and the failure reason. Then match it against Usage History on the Background Agents page.
- An "Out of credits" row means the workspace balance ran out. Agents are paused and admins are emailed.
- A "Daily limit exceeded" row means the attempt went past the agent's daily run limit. It is logged for the audit trail, no credits were charged, and the agent was not deactivated.
- "No messages in this run" means the run produced no conversation, usually because it failed very early, for example on the credit check.
- A tile showing a dash instead of a number, labelled "Runs unavailable" or "Usage unavailable", means the request behind it failed. Spiich refuses to print a number that would be read as zero. Reload, or use
Try againin Recent runs.
Two things that are not faults
A scheduled agent that runs at the wrong time of day is running in the timezone of the member it runs as, not yours, and a member with no timezone set falls back to UTC. Set the member's timezone at Settings > User Settings, or move the schedule. The form says "Times are in the agent member's timezone" and the list shows the resolved next run.
Greyed-out pause, edit and delete buttons on somebody else's agent are also intentional: the agent runs as an admin, so editing it would change what an admin's identity does on the next run. Ask an admin, or click the copy icon to make your own version that runs as you.
Still stuck
Email hello@spiich.ai with the agent name, the run you are asking about, and the status and failure reason from its header. We can read the run from our side. You can also use the contact form.