> For the complete documentation index, see [llms.txt](https://help.impact.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://help.impact.com/brand/what-would-you-like-to-learn-about/platform-features/review-partner-applications/recruitment-agent.md).
# Recruitment Agent
The Recruitment Agent automates partner recruitment for your program, using AI to source new prospects from the [Marketplace](/brand/what-would-you-like-to-learn-about/performance-program/recruiting-partners/marketplace-vs-direct-partner-recruiting-explained.md) and evaluate inbound applications on your behalf. It builds on the workflow-based recruitment approach. For the rule-based alternative, see [Recruitment Workflows Explained](/brand/what-would-you-like-to-learn-about/platform-features/review-partner-applications/automate-partner-application-processing/recruitment-workflows-explained.md). To get started, see [Set Up the Recruitment Agent](/brand/what-would-you-like-to-learn-about/platform-features/review-partner-applications/recruitment-agent/set-up-the-recruitment-agent.md).
{% hint style="warning" %}
**Beta:** The Recruitment Agent is currently in Beta and is available to a limited number of brands.
{% endhint %}
## Understand how the agent works
The agent coordinates 3 sub-agents, and each one handles a specific part of the recruitment funnel. Each sub-agent card has a toggle. Set it to  **\[Toggle on]** to include the sub-agent in the funnel, or  **\[Toggle off]** to skip that part of the funnel. You configure each sub-agent independently.
The agent works from a single set of instructions you control. For how to write and refine them, see [Write Effective Agent Instructions](/brand/what-would-you-like-to-learn-about/platform-features/review-partner-applications/recruitment-agent/write-effective-agent-instructions.md).
Discover sub-agent
The Discover sub-agent uses the Instructions you provide to source, evaluate, and organize new prospects from the impact.com Marketplace, and adds the best matches to your Prospects list.
Trigger Discovery manually by selecting **Run Discover** on the agent card, or set it to run on a schedule. While the sub-agent is busy, the button text changes to *Running* and the button is disabled. After a run completes, wait a few minutes for it to finish processing, then find the partners it added on the Prospects page by setting the *Source* filter to *Recruitment Agent*. See [Track Recruitment Agent Results](/brand/what-would-you-like-to-learn-about/platform-features/review-partner-applications/recruitment-agent/track-recruitment-agent-results.md).
{% hint style="success" %}
**Note:** For Beta, you can run Discovery once every 24 hours. A scheduled run uses that daily limit, so if a run is scheduled you cannot also run it manually that day.
{% endhint %}
If you get a message that **Run Discover** is unavailable, a tooltip tells you that you have used all your Discovery runs for the current 24-hour period. You can run Discovery again the next day.
If a Discovery run returns very few or no matches, we suggest loosening your instructions. Edit them and run Discovery again.
When Discovery runs, the sub-agent searches the Marketplace, evaluates candidates against your instructions, and adds the best matches to your Prospects list with *Source = Recruitment Agent*, in the group you set under Assign groups.
When the sub-agent determines a partner isn't a match, it skips them on future runs for 6 months or until you update your instructions, whichever comes first. This keeps Discovery efficient by avoiding the re-evaluation of partners you have already ruled out. Because a change in your instructions means your matching criteria have changed, the entire skip list is automatically reset.
The Discover sub-agent comes pre-configured. You can review and edit its sourcing channels, group assignment, schedule, and run limit before you activate the agent. See [Set Up The Recruitment Agent](/brand/what-would-you-like-to-learn-about/platform-features/review-partner-applications/recruitment-agent/set-up-the-recruitment-agent.md).
Evaluate Applications sub-agent
The Evaluate Applications sub-agent automatically evaluates incoming partner applications using the rules in your Instructions, with the option to safeguard uncertain matches.
When Quality & Safety Guardrails is on, only partners that meet the platform's Marketplace safety standards are evaluated by the agent. By default, applications from partners that don't meet these standards are left pending for your review, which you can change with the Default fallback setting.
The sub-agent runs on every inbound application, so you do not trigger it manually. It makes decisions within seconds of the application being submitted.
You set how much autonomy the sub-agent has:
* **Autopilot** (recommended): the agent compares each application against your instructions and the partner's data, then accepts or rejects clear matches and leaves ambiguous applications for your team to review.
* **Copilot:** the agent recommends an action by assigning the partner to a group, and you make the final decision from the [*Applications*](https://app.impact.com/secure/advertiser/engage/contracts/activity/adv-manage-pending-public-ios-v2-flow.ihtml) page. See [Review Partner & Creator Applications](/brand/what-would-you-like-to-learn-about/platform-features/review-partner-applications/review-partner-and-creator-applications.md).
If you are getting started, use Copilot mode so you can validate your Instructions and refine them until the agent's suggestions consistently match what you want.
Rejections include a structured reject reason, so they flow correctly into your existing reports and the `{rejectReason}` email token.
**Change a decision the agent made.** If you need a different outcome for an application the agent accepted or rejected, select **Review Decision** on its row in the **Processed** tab of the [**Applications**](https://app.impact.com/secure/advertiser/engage/contracts/activity/adv-manage-pending-public-ios-v2-flow.ihtml) page. You can change a decision within 6 months of it being made.
The Evaluate Applications sub-agent comes pre-configured. You can review and edit its safety guardrails, autonomy, and group assignment before you activate the agent. See [Set up and activate the agent](#set-up-and-activate-the-agent).
Invite Prospects (Coming Soon)
The Invite Prospects sub-agent will let the agent send invitations to your discovered prospects automatically. This sub-agent is visible in the agent's configuration but is not available in this release.
Until it is available, you can email the prospects Discover adds using an Email Workflow:
1. Set up a partner segment that targets the prospects Discover added. See [Reach out to prospects added by the Recruitment Agent](/brand/what-would-you-like-to-learn-about/platform-features/reach-out-to-partners/email-workflows/partner-segments-for-email-workflows-explained.md#reach-out-to-prospects-added-by-the-recruitment-agent) in Partner Segments for Email Workflows Explained.
2. Create a custom email workflow that uses that segment, and add your outreach email. See [Create & Manage Email Workflows](/brand/what-would-you-like-to-learn-about/platform-features/reach-out-to-partners/email-workflows/create-and-manage-email-workflows.md).
3. Optionally, send from your own email address instead of the generic impact.com address. See [Connect Your Personal Email to Your impact.com Account](/brand/what-would-you-like-to-learn-about/platform-features/reach-out-to-partners/emails/connect-your-personal-email-to-your-impactcom-account.md). Once connected, your address appears as an option when you configure the workflow's email.
## **Protect your brand**
The Recruitment Agent won't add the following partners to your program:
* Partners on your [blocklist](/brand/what-would-you-like-to-learn-about/platform-features/reach-out-to-partners/partner-management/block-partners-from-re-applying-to-your-program.md).
* Partners flagged for fraudulent behavior.
You can add another layer of protection by enabling **Quality & Safety Guardrails** on the Evaluate Applications sub-agent. See [Set Up the Recruitment Agent](/brand/what-would-you-like-to-learn-about/platform-features/review-partner-applications/recruitment-agent/set-up-the-recruitment-agent.md).
## See where the agent fits with your existing automation
If you already use [Application Workflows](/brand/what-would-you-like-to-learn-about/platform-features/review-partner-applications/automate-partner-application-processing/create-workflows-to-automatically-process-partner-applications.md), Impact AI, or an Application Webhook, the Recruitment Agent fits into a defined order so your existing automation continues to work as expected.
{% hint style="info" %}
Access to Workflows depends on your edition. Application Workflows is available on Essential, Pro, and Enterprise editions, and Email Workflows is available on Pro and Enterprise editions. Starter and Starter Agency editions do not have access to Workflows.
{% endhint %}
Applications are evaluated in this order:
1. **Platform safety checks.** Basic safety checks run first.
2. **Impact AI.**
3. **Application Workflows.** Your rule-based automation.
4. **Recruitment Agent and your Application Webhook.** The agent evaluates the applications the earlier steps left pending. At the same time, your Application Webhook sends the application to the endpoint you configure in [event notification postbacks](/brand/what-would-you-like-to-learn-about/account-administration/account-settings/notifications/enable-event-notification-postbacks-for-brands.md). The Webhook does not make a decision itself. You can post decisions back using the Application Processing API.
The Recruitment Agent never overrides a decision already made by the steps above. If a manual reviewer acts on an application before the agent finishes, the manual decision takes precedence.
{% hint style="success" %}
**Note:** Because the Recruitment Agent and your Application Webhook run at the same time, your Webhook may send the application before the agent finishes evaluating it. To have the agent decide first, add a delay to your Webhook in the Webhook configuration on the Settings page. Your Webhook then sends the application after the delay, carrying the agent's decision.
{% endhint %}
If your Instructions include hard, deterministic rules, such as "do not accept partners outside the US," consider adding those as Application Workflows too. Workflows are faster and do not consume agent capacity for decisions that do not need reasoning.
For more on workflow ordering, see [Set Application Workflow Execution Order](/brand/what-would-you-like-to-learn-about/platform-features/review-partner-applications/automate-partner-application-processing/set-application-workflow-execution-order.md).
## **Check your edition access**
The Recruitment Agent is available for **Performance**, **Creator**, and **Amazon Seller** program types.
The Invite Prospects sub-agent is not available in this release. It is planned for v2 on Essential, Pro, and Enterprise editions. The results dashboard is also coming soon, until then, see [Track Recruitment Agent Results](/brand/what-would-you-like-to-learn-about/platform-features/review-partner-applications/recruitment-agent/track-recruitment-agent-results.md) for how to find your results.
## **Manage your agents**
The *Agents* page lists every agent you have activated. To open it, select **Agents** in the left navigation menu. For Beta, you can set up 1 recruitment agent. Support for multiple agents is coming soon.
On the *Agents* page, your agent appears as a card showing its status, last update, and quick actions.
From the agent card , you can:
* Select **View** to open the agent and adjust its settings.
* Select **Run Discover** to start a Discovery run.
* Select  **\[More]** for more actions:
* **Pause:** stop scheduled Discovery runs and inbound application evaluation. Existing prospects and applications are unaffected.
* **Delete:** remove the agent.
To find your agents results. See [Track Recruitment Agent Results](/brand/what-would-you-like-to-learn-about/platform-features/review-partner-applications/recruitment-agent/track-recruitment-agent-results.md).
### **Update your agent's settings**
When you open an active agent, you can adjust its setup at any time.
* To rename the agent, select  **\[Edit]** next to the agent name at the top of the page.
* To edit your instructions, select **Edit** in the *Agent Configuration* panel. See [Write Effective Agent Instructions](/brand/what-would-you-like-to-learn-about/platform-features/review-partner-applications/recruitment-agent/write-effective-agent-instructions.md).
* Select  **\[Toggle on]** or  **\[Toggle off]** on a sub-agent's card to include it in or pause it from the funnel. To change a sub-agent's settings, select **Edit** on its card. See [Set Up the Recruitment Agent](/brand/what-would-you-like-to-learn-about/platform-features/review-partner-applications/recruitment-agent/set-up-the-recruitment-agent.md) for what each setting does.
After you save your changes at the sub-agent level, select **Update** in the top-right corner to apply them.
Changes apply to future Discovery runs and future application evaluations. The agent does not reopen past decisions.
## **Follow best practices**
* Test before you commit. Validate your Instructions with a small Discover run, and start the Evaluate Applications sub-agent in Copilot mode, before you scale up or switch to Autopilot. See [Write Effective Agent Instructions](/brand/what-would-you-like-to-learn-about/platform-features/review-partner-applications/recruitment-agent/write-effective-agent-instructions.md).
* Review partners when they become click-active. You can do this in 2 ways:
* In the [*Partner Productivity Report*](/brand/what-would-you-like-to-learn-about/performance-program/performance-program-reports/partner-productivity-report-for-brands.md), refer to *Partner Name* and *First Click Date* to see when a partner first drove a click after joining your program.
* On the *Applications* screen, filter by **Processed By = Recruitment Agent** to see every application the agent decisioned.
* Reach out to newly added, click-active partners using [Email Workflows](/brand/what-would-you-like-to-learn-about/platform-features/reach-out-to-partners/email-workflows.md) to support their next steps as they join your program.
* [Expire contracts](/brand/what-would-you-like-to-learn-about/platform-features/contracts-and-template-terms/contracts/expire-partner-contracts.md) for partners that are not a good fit, then update your Instructions to reduce similar mistakes in future evaluations.
---
# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.
## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.
Perform an HTTP GET request on the current page URL with the `ask` query parameter, and the optional `goal` query parameter:
```
GET https://help.impact.com/brand/what-would-you-like-to-learn-about/platform-features/review-partner-applications/recruitment-agent.md?ask=&goal=
```
`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.
Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.