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

# Start from the Catalogue

> Find a ready-made use case in the Catalogue, activate it as a draft agent on your own documents, publish it and run it.

The Catalogue is a library of ready-made agents and workflows for common office jobs, which you copy and adapt instead of building from scratch. Here you turn its "Policy Q\&A Agent" into an agent of your own that answers questions from your policy wordings, then publish it and ask it something.

> [!NOTE]
>
> **Time:** about 20 minutes · **Role:** Member or Admin (Viewers have no **Catalogue**) · **Works on:** Enterprise and SaaS. On the SaaS Free plan you can have 3 workers (agents and workflows together) and 1 knowledge base. · **You need:** two or three policy wordings (PDF or Word), for example "Home insurance policy wording" and "Travel insurance policy wording"

<video controls preload="metadata" poster="/videos/start-from-the-catalogue.jpg" src="https://mintcdn.com/neuralfactory/y-MI7aIFykaw0cY2/videos/start-from-the-catalogue.mp4?fit=max&auto=format&n=y-MI7aIFykaw0cY2&q=85&s=c4c03bddbf508774d3f592bc2df9ee66" />

Watch: Start from the Catalogue (1 min 13 s)

## What you will have at the end

* A published agent made from a Catalogue use case, bound to your own knowledge base.
* That knowledge base, with your wordings processed.
* One task in which the agent answered from those wordings.

## 1. Browse the Catalogue and open a use case

Activating a use case gives you a copy you can change.

**1. Open the Catalogue**

In the sidebar, under **Workforce**, click **Catalogue**.

**2. Narrow the list**

Under **What it is**, tick **AI agent**. Type "policy" in **Search the catalogue**.

**3. Open the use case**

Click **Policy Q\&A Agent**. Read **Who it is for**, **The problem**, **What it does** and **How it runs** on the left. Some use cases also list **What it does not do**.

**4. Look at Readiness**

On the right, **Readiness** lists what the use case needs. The Policy Q\&A Agent needs no toolkit, only one set of documents, "Policy Wordings", marked **You provide**; when a knowledge base of that name already exists, the row says **Already in your knowledge bases** instead. The badge reads **Ready to activate**. Other use cases read **One connection away** or **Needs your system** when they reach for a toolkit or one of your own systems. You can activate those too: the agent is created without the toolkits your organization has no connection for.

![The Policy Q\&A Agent use case, with Who it is for, The problem, What it does and How it runs on the left, the Readiness card with the Ready to activate badge on the right, and the Activate button at the top right.](https://mintcdn.com/neuralfactory/y-MI7aIFykaw0cY2/images/app/catalogue-use-case.png?fit=max\&auto=format\&n=y-MI7aIFykaw0cY2\&q=85\&s=032f692dd40f692f35b0f555aa2e92e1)

> [!TIP]
>
> The page shows an **Activate** button. If it is greyed out, see [If something goes wrong](#if-something-goes-wrong).

## 2. Activate it

**1. Click Activate**

Four steps follow: **Connect**, **Documents**, **How it runs** and **Build**. Move between them with **Back** and **Continue**.

**2. Connect**

**Connect what it reaches for** says **Everything is connected. Nothing to do here.** Click **Continue**. A use case that reaches for toolkits lists them here; you can continue without connecting them and connect them later, see [Connect an agent to your tools](https://docs.neuralfactory.ai/user-guide/guides/connect-your-tools.md).

**3. Documents**

Under **Point it at your documents**, choose the knowledge base for "Policy Wordings". If your wordings are already in a knowledge base, pick it; one named "Policy Wordings" is picked for you. Otherwise keep **Create “Policy Wordings”** to get a new, empty knowledge base. On the SaaS Free plan, which allows one knowledge base, pick the one you already have if you have one. Click **Continue**.

**4. How it runs**

Leave **How it starts** on **In a conversation**. This use case has **No approval gate**. Under **Who can use it**, keep **Only me, for now** while you try it. Click **Build it**.

**5. Build**

The step creates the knowledge base if you asked for a new one, then the agent. It ends on **Your draft is ready**.

> [!TIP]
>
> When something is left to do, **Before you publish** names it. If you created a new knowledge base, expect a line saying that "Policy Wordings" has no documents yet.

## 3. Add your documents

Skip this section if you picked a knowledge base that already holds your wordings.

**1. Open the knowledge base**

Under **Knowledge**, click **Knowledge bases**, then "Policy Wordings" (or the knowledge base you picked).

**2. Add a File data source**

Click **Add Source**, choose **File**, type a **Data Source Name** and a **Description**, and click **Create**.

**3. Upload and wait**

Click the new source in **Data Sources**, then **Upload**, and choose your files. Wait until each one shows **Processed**.

> [!TIP]
>
> "Policy Wordings" shows **Indexed** in the **Knowledge bases** list.

## 4. Publish and run it

Nobody can start a task with a draft until you publish it.

**1. Open the draft**

Click **Open the draft** on the last activation step, or go to **Workforce** › **Agents** and click "Policy Q\&A Agent", which has the **Draft** badge. It opens in the studio.

**2. Review it**

Read **Instructions** and **Guardrails**. The agent answers only from its knowledge base, quotes the clause it relies on and cites its sources. Change anything with **Edit**.

**3. Publish**

Click **Publish** in the header. Write who reviewed it under **Review note (optional)** and click **Publish**.

**4. Ask it something**

Click **New task** in the studio's header. Ask a question your wordings answer, such as "What does our home insurance policy, version 2026.1, say about a burst pipe?", and press **Enter**.

> [!TIP]
>
> The answer quotes the clause from your wording, with a numbered citation such as \[1], and **1 source used** below it. Click **1 source used** to see the file it came from.

When it works, share it from **Who can use it** in the studio, then publish again (Enterprise, or a SaaS Business organization). See [Agent access control](https://docs.neuralfactory.ai/user-guide/agents/agent-access-control.md).

## If something goes wrong

* **Activate is greyed out** (SaaS): your plan has no room for another worker. The tooltip says how many workers (agents and workflows) the use case needs and how many you use. Delete an agent or workflow you no longer need, or upgrade. See [Plans and limits](https://docs.neuralfactory.ai/user-guide/admin/billing.md).
* **The Build step shows It could not be built**: the reason follows. Click **Try again**: it repeats only what failed, and reuses a knowledge base an earlier attempt created.
* **The agent asks which product and policy version you mean**: it does this whenever a question could apply to more than one wording. Answer it, or name the product and version in the question.
* **The agent says the policy text does not specify this**: check that your documents are **Processed** and that they answer the question. The agent never answers from general knowledge.
* **Connect shows Ask an administrator** (Enterprise, for a use case that needs a toolkit): only Admins connect toolkits there. Continue without it, or ask an Admin.

## Where to go next

* [Catalogue overview](https://docs.neuralfactory.ai/user-guide/catalogue/overview.md)
* [Activating a use case](https://docs.neuralfactory.ai/user-guide/catalogue/activating-a-use-case.md): every option of the four steps
* [Build your first agent](https://docs.neuralfactory.ai/user-guide/guides/first-agent.md)


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.