Agents run tasks, such as translation, summarization, and SEO analysis, on content entries. You can trigger them from the content table (on one or more entries at once) or from the content entry form (on a single entry). When an agent runs, it writes its output directly to the entry. The entry then enters a review state so you can check the changes and approve or revert changes field by field.
For an overview of what agents do and how they work, see AI Agents.
#Before you begin
Hygraph provides three agent types. Your admin configures which ones are available in your project and which models they apply to.
| Agent | What it does |
|---|
| SEO Expert | Reviews your entry and posts a structured SEO report as a comment, covering content optimization, heading structure, and metadata gaps. |
| Translator agent | Translates content fields into one or more target locales. |
| Content summarizer | Generates a concise summary from one or more fields and writes it to a target field. |
#Trigger an agent
Agents are available on entries in models that an admin has configured for agent use. If you don't see an agent option in the content table or content entry form sidebar, your admin needs to configure the agent for that model.
When the Translator agent runs, it overwrites all existing content for the selected locales, including fields that have not changed. Review the localized content after each run to ensure no previously translated fields were unintentionally overwritten.
#Bulk updates in the content table
Use this option to run an agent on multiple entries at once.
- Go to the Content tab and open a view of a content model.
- Select one or more entries using the checkboxes in the content table.
- In the action bar at the bottom of the screen, click Agents and select the agent you want to run.
The agent runs on all selected entries, up to a maximum of 50 entries per run. The selected content entries become read-only while the agent runs.
#Single entry updates in the content entry form
Use this to run an agent on a single entry you have open.
- Open the entry in the content editor.
- In the right sidebar, locate the Agents panel.
- Select the agent you want to run and confirm.
The content entry form becomes read-only immediately while the agent runs.
#Configure a Content summarizer agent run
When you trigger a Content summarizer, a configuration panel appears before the run starts.
- Under Input fields, select the source fields containing the content to summarize.
- Under Entries, review the content entries whose source fields will be summarized. This field is read-only.
- Under Output field, select the field where the generated summary will be written.
- Under Tone, choose the tone for the generated summary.
- Confirm to start the run.
For bulk runs from the content table, the same configuration applies to all selected entries.
#What happens during a run
Once triggered, the agent processes the entry and writes its output directly to the relevant fields. While this is happening:
- The content entry form is read-only. No edits can be made to the entry.
- The entry shows a Pending status in the content table.
When the run completes, the agent's changes are in place and the entry status changes to Ready for review. The entry stays read-only until you complete the review.
If the run does not complete successfully, the entry shows a Failure status and the entry content is unchanged.
#Review agent output
The content table shows agent status so you can find entries that need action without opening each one individually.
| Agent status | Description |
|---|
| Pending | Entries where an agent run is queued or in progress |
| Ready for review | Entries where a completed run is awaiting your review |
| Failure | Entries where the agent run did not complete successfully |
When an entry shows Ready for review, open it to check what the agent changed and approve or revert changes field by field. Each content entry remains read-only until the review is complete.
- Open the entry. The review screen shows a side-by-side diff: the original content on one side, the agent's output on the other.
- Review each field the agent changed.
- Approve changes you want to keep, or revert any you don't.
- If you revert all fields, the entry returns to its exact pre-agent state.
- Once you have finished reviewing, the entry exits read-only state and is editable again.
The agent writes to the draft version of your content only. Nothing reaches your published content until you publish the entry.
You cannot edit an entry while a run is in progress or while a completed run is awaiting review. Complete the review to restore edit access.
#Guidelines and agent output
If an agent is assigned guidelines by your admin, its output reflects your project's brand voice, terminology, audience requirements, or compliance constraints. You won't see the guidelines themselves during a run, but their effect shows up in the output.
If the output doesn't match your expectations, ask your admin to review the guidelines assigned to that agent. See AI Guidelines for what each category covers.
#Troubleshooting
The agent may not be configured for that model. Contact your admin to confirm the agent is set up for Use in Agent runs and that your model is included in its scope.
#Entry is stuck in read-only state
The entry has a completed run awaiting review. Open the entry, complete the review by approving or reverting each field, and the entry will exit read-only state. If you did not trigger the run yourself, another user may have triggered it. Check the Ready for review status in the Content Table.
#All selected entries could not be triggered
This can happen for two reasons:
- Every selected entry has an active run in progress or a completed run awaiting review. Complete the pending reviews first, then trigger the agent again.
- A selected entry has an editor currently active on it. Wait until the entries are no longer being edited, then trigger the agent again.
#No fields available to select in Content Summarizer configuration panel
For Input fields, the entry needs at least one supported field with existing content. Confirm the entry has content in the fields you want to summarize.
#Output does not match expectations
The agent may not have guidelines assigned, or the existing guidelines may need updating. Ask your admin to review the guidelines assigned to that agent. See AI Guidelines for what each category covers.
#What's next
- AI tools for editors: Overview of all AI tools available to editors.
- Use AI Assist: Run AI Assist tasks on individual fields within an entry.
- AI Guidelines: Understand what guidelines are and how to request changes from your admin.