---
title: "Review agent readiness before launch"
description: "Use Interlinked readiness checks to find what an AI agent still needs before it can be published, activated, and assigned to customers."
section: "AI agent management"
url: https://interlinked-ai.com/en/resources/docs/review-agent-readiness
lang: en
lastUpdated: 2026-08-13
reviewedAt: 2026-08-13
---

# Review agent readiness before launch

## What readiness means

Readiness is the panel’s checklist for whether an agent has the information and connections needed to work safely. It is separate from the agent’s status. A draft can be well configured but not live; an active agent can later become unavailable if a channel or required connection changes.

Open **My Agents** to see a readiness count on each card, or open the agent to see **Agent readiness** and the specific blockers.

<div class="docs-media-placeholder">
  <p class="mono-label">Screenshot: Agent readiness panel showing completed checks, a blocker, and the action to fix it</p>
</div>

## Read the checklist

The panel distinguishes required items from recommendations and optional information. Common checks include:

- **Goal defined** — The agent has a clear purpose.
- **AI model selected** — A model is configured for the agent.
- **App connected** — At least one communication app is connected, unless the agent is internal-only.
- **Human handoff configured** — The agent has a path for handing work to your team when needed.
- **Product catalog ready** — Required for workflows that rely on products.
- **Calendar connected** — Required for appointment journeys that use scheduling.
- **Capabilities selected** — The agent has at least one supported ability when the workflow needs it.
- **Safety rules configured** — The agent has the default or selected safety rules.
- **Test conversation completed** — The panel may recommend testing before launch.

An item can be **Required**, **Recommended**, or **Optional**. A required item blocks activation until it is complete. Select the action beside the item to open the relevant panel area.

## Fix the blocker the panel names

Read the exact message before changing settings. Typical launch blockers include:

| Panel message | What to check |
| --- | --- |
| **Publish a valid agent version** | Review the draft and publish it from **Unpublished changes**. |
| **Assign and activate a customer channel** | Connect a channel, assign it to this agent, and confirm the channel is active. |
| **Setup required before turning on** | Open the agent and follow each linked requirement. |
| **Configure an AI model** | Open the model area and choose an available model. |
| **Add products to your catalog** | Add or import products if the workflow sells products. |
| **Connect your Google Calendar** | Connect the calendar before activating an appointment service. |

The exact list depends on the workflow and whether the agent is internal or customer-facing. A requirement that does not apply may be labeled **Not applicable** or **Internal mode — no external app required**.

## Readiness is not activation

When the panel reports **Ready to activate**, the agent has passed the checks currently visible to you. It still does not answer customers until you intentionally use the activation control. Likewise, publishing a draft changes the live configuration but does not automatically turn an agent on or off.

This separation lets you publish a reviewed configuration while keeping the agent paused, or fix a missing channel before activating it. See [Publish and activate an agent](/en/resources/docs/publish-and-activate-an-agent) for the complete sequence.

## Recheck after important changes

Review readiness again after:

- changing the workflow or customer-facing mode;
- assigning a channel to a different agent;
- disconnecting or reconnecting WhatsApp, Instagram, Messenger, or Google Calendar;
- changing product or service requirements; or
- enabling an ability or governance checkpoint.

If a check looks stale, refresh the agent page and follow the linked action. If the panel still reports a blocker after the requirement is visibly complete, use [Troubleshooting and support](/en/resources/docs/troubleshooting-and-support-overview) and include the agent name and the exact readiness message.