> ## Documentation Index
> Fetch the complete documentation index at: https://docs.hyperline.co/llms.txt
> Use this file to discover all available pages before exploring further.

# Customer monitoring agent

> Configure the Hyperline customer monitoring agent to analyze account health, surface risk signals, and notify the right teammates automatically.

The customer monitoring agent analyzes selected customers and produces a health score with supporting context. It helps you spot accounts that need attention because of payment issues, usage changes, support signals, contract risk, weak communication, or expansion potential.

The agent benefits from Hyperline's revenue management context, including contracts, subscriptions, invoices, payments, and usage. This gives the health score a revenue-aware foundation before it is enriched with connected tools such as Gmail, support platforms, meeting tools, and CRM systems.

You can use one monitoring agent for your full customer base, or focus it on a specific [customer segment](../customers/segments) such as enterprise customers, customers near renewal, or customers with late payment behavior.

## Where to find it

Go to **Customers > Monitoring** to create and manage monitoring agents.

After an agent has analyzed a customer, the customer page can show:

* A health score card on the customer overview
* The last monitoring date
* A link to the latest analysis
* The monitoring agent assigned to that customer

## What the agent analyzes

The agent combines billing and customer signals across several dimensions:

| Dimension              | What it considers                                                    |
| ---------------------- | -------------------------------------------------------------------- |
| Payment health         | Payment consistency, failed payments, disputes, and overdue invoices |
| Usage and engagement   | Product usage trends and declining activity                          |
| Support signals        | Recent ticket volume, sentiment, and frustration patterns            |
| Contract and valuation | Contract terms, renewal risk, and revenue at stake                   |
| Communication          | Meeting activity, email engagement, and relationship quality         |

Connected tools can add more context. For example, Gmail can add recent customer email threads, support tools can add ticket history, meeting tools can add summaries and transcripts, and CRM tools can add account records, deals, contacts, and notes.

These dimensions are analysis angles, not a rigid checklist. Depending on the customer context, the agent can emphasize the most relevant angle, such as payment risk, billing issues, expansion readiness, renewal risk, weak communication, or unresolved support issues.

## How context is used in analysis

During each run, Hyperline pulls the relevant context for each monitored customer from Hyperline and connected tools. The agent uses that context to explain the health score, not just to list recent activity.

This helps the analysis connect signals that are often reviewed separately:

* A late invoice can be interpreted with payment history, recent emails, open support threads, and renewal context.
* A usage increase can be checked against CRM deal context and support sentiment to surface possible expansion readiness.
* A usage drop can be compared with recent meetings, email silence, or unresolved support issues to identify churn risk.
* A high-value customer can be prioritized differently when contract size, payment behavior, and relationship activity point in different directions.

<Card title="Example 💡">
  A customer has growing usage and an upcoming renewal, but recent Intercom conversations show unresolved implementation blockers. The agent can surface both the expansion signal and the risk, so the owner can prepare the right follow-up before starting a commercial conversation.
</Card>

## Use cases

| Use case                    | How the agent helps                                                                          |
| --------------------------- | -------------------------------------------------------------------------------------------- |
| Upsell and expansion review | Surfaces customers with strong usage, positive signals, and meaningful contract value        |
| Churn risk review           | Highlights declining usage, weak engagement, negative support signals, or payment issues     |
| Renewal preparation         | Summarizes health, risks, positive signals, and recent activity before renewal conversations |
| Collections follow-up       | Adds payment health and late invoice context to account-level monitoring                     |
| Onboarding monitoring       | Helps identify customers that have not reached expected usage or engagement milestones       |
| Executive account review    | Gives account owners a concise health summary with supporting reasoning                      |

## Choose a setup mode

When you create a monitoring agent, choose how much control you want over the analysis.

| Mode          | Best for                                                                           | How it works                                                                                  |
| ------------- | ---------------------------------------------------------------------------------- | --------------------------------------------------------------------------------------------- |
| **Autopilot** | Quick setup or broad customer monitoring                                           | Hyperline chooses the schedule and signal weighting based on billing data and connected tools |
| **Advanced**  | Specific monitoring playbooks, such as collections, renewals, or expansion reviews | Customer scope, schedule, fine tuning, context, connectors, and notifications                 |

Use **Autopilot** when you want the agent to start monitoring with minimal setup. Use **Advanced** when you want to adapt the agent to a specific team workflow or customer segment.

## Create a monitoring agent

1. Go to **Customers > Monitoring**.
2. Click **Create agent**.
3. Open the agent and go to the **Configuration** tab.
4. Choose **Autopilot** or **Advanced**.
5. Choose which customers the agent monitors:
   * **All customers**
   * A specific customer segment
6. Click **Save changes**.

When you change the monitored customer scope, Hyperline asks you to confirm the assignment changes. This helps prevent accidental changes when a segment includes a large number of customers.

<Tip>
  Use segments to keep an agent focused. For example, monitor "Enterprise customers renewing in the next 90 days" separately from lower-touch accounts.
</Tip>

## Fine tune the agent

In **Advanced** mode, use the configuration tabs to customize how the agent works:

| Setting            | What it controls                                                               |
| ------------------ | ------------------------------------------------------------------------------ |
| **Customer scope** | Whether the agent monitors all customers or a selected customer segment        |
| **Schedule**       | Whether the agent runs automatically or on a fixed schedule                    |
| **Fine tuning**    | How strongly each signal dimension contributes to the health score             |
| **Context**        | Instructions the agent should use when interpreting customer signals           |
| **Connectors**     | Connected tools the agent can use for CRM, support, meeting, or custom context |
| **Notifications**  | Who the agent can ping and which findings should trigger a notification        |

### Configure schedule and scoring

In **Advanced** mode, use **Schedule** to decide when the agent runs:

* **Automatic**: Hyperline monitors customers when relevant events happen, such as new emails, late invoices, or usage changes.
* **Fixed schedule**: the agent runs at a regular interval that you define.

Use **Fine tuning** to adjust how strongly each dimension contributes to the health score. For example, you can give more importance to **Payment health** for collections-focused monitoring, to **Usage and engagement** for churn-risk monitoring, or to **Contract and valuation** for expansion and renewal reviews.

### Add context

Use the **Context** tab to give the agent additional instructions:

* **General context**: company or customer-base context the agent should know.
* **Scoring instructions**: guidance for interpreting signals, such as industry-specific thresholds or risk rules.

Keep context concise and operational. The best instructions explain how your team interprets customer risk, not generic business goals.

## Enable connectors

Use the **Connectors** tab to choose which connected tools the agent can use. The tab can include CRM, support, meeting, and custom MCP data sources.

Each connected data source has a toggle. Unavailable connectors appear disabled with **Connect to enable** until the underlying integration is connected.

Gmail email context is handled separately. When **Gmail** is connected for users in your workspace, the agent can use recent customer email threads for matching customer domains. Gmail can appear in **Data sources** on an analysis result even though it is not toggled from the agent **Connectors** tab.

| Data source    | Examples                   | What it adds                                                                                                                          |
| -------------- | -------------------------- | ------------------------------------------------------------------------------------------------------------------------------------- |
| Email          | Gmail                      | Recent customer email threads, including subjects, participants, dates, and thread details when the agent needs communication context |
| Support        | Plain, Intercom, Zendesk   | Support threads, statuses, priorities, labels, recent messages, and unresolved customer issues                                        |
| Meetings       | Claap                      | Meeting records, participants, summaries, and transcripts when available                                                              |
| CRM            | Attio, HubSpot, Salesforce | Company records, contacts, deals, CRM meetings, and CRM notes                                                                         |
| Custom sources | MCP connections            | Additional customer context from internal tools or custom data sources connected to the agent                                         |

<Tip>
  Keep customer **Domain** and **Billing email** values up to date. They help Hyperline match Gmail messages, support threads, and meetings to the right customer.
</Tip>

<Card title="Example 💡">
  Before a renewal review, the agent can combine Hyperline billing data with Gmail threads, Intercom support conversations, Claap meeting summaries, and HubSpot deal context. The result helps the owner see payment risk, open support issues, recent engagement, and expansion signals in one customer health analysis.
</Card>

## Configure notifications

Use the **Notifications** tab to decide when and where the agent can ping your team.

You can configure:

* **Customer owner**: ping the user responsible for the monitored customer.
* **Customer followers**: ping users following the customer.
* **Specific people**: always ping selected users.
* **Noise control**: only notify for findings at or above a selected severity.
* **Slack threads**: post into a Slack channel when Slack is connected.

For more about customer ownership and followers, see [Owner and followers](./owner-followers).

<Note>
  Agent notification settings decide which recipients the agent may ping. Each recipient can also control their own channels and customer event subscriptions in **Profile > Notifications**.
</Note>

## Run the agent manually

You can trigger a manual run from the agent page, or from a customer health score card when a customer has not been monitored yet.

When triggering a run, you can:

1. Select one or more customers.
2. Adjust scoring weights for that run.
3. Add one-time context for the analysis.
4. Start the run and open the run detail page.

Manual runs are useful before a customer meeting, upsell review, renewal review, collections follow-up, or board reporting cycle.

## Read analysis results

Each completed analysis can include:

* Overall health score
* Score tier
* Short description
* Reasoning
* Dimension scores
* Positive signals
* Risk factors
* Summary
* Data sources and tools used

| Score range | Tier            |
| ----------- | --------------- |
| `0-25`      | Critical        |
| `26-50`     | At risk         |
| `51-75`     | Needs attention |
| `76-100`    | Healthy         |

<Note>
  A health score is a prioritization signal. Use the reasoning, risk factors, and source data to decide what action to take.
</Note>

## Related

* [Customer intelligence overview](./overview)
* [Owner and followers](./owner-followers)
* [Custom usage tracking](./custom-usage-tracking)
* [Late payment behavior](./late-payment-behavior)