Agents

Create an AI agent

An agent is your branded AI assistant. You give it a name and some knowledge, and Ultimo Bots turns that into a working assistant that can answer your visitors questions. Most agents are ready in under a minute.

You create agents from the Your Agents screen, which is the first page you land on after signing in to the dashboard. Each agent is independent: it has its own knowledge, design, behavior, and analytics.

An agent answers only from the knowledge you give it. You can create an agent with no knowledge to set it up first, but it will not be able to answer questions until you add a website or documents.

Before you start

You only need one of the following to give your agent something to talk about:

A public website URL. Ultimo Bots crawls the site and learns from the pages it can reach.
Documents. Upload files such as PDFs or Word documents (handbooks, FAQs, product sheets).

The agent name is the only required field. The website and documents are both optional, and you can add or change knowledge at any time after the agent is created.

Open the New Agent dialog

On the Your Agents screen, click New Agent. The button appears both at the top of the agent list and at the bottom of the left sidebar.

The Your Agents screen with existing agents and the New Agent button — The Your Agents screen. Click New Agent to open the creation dialog.

Fill in the form

The dialog has three sections. Only the first is required, so the fastest way to create an agent is to enter a name, paste your website URL, and continue.

Name your agent
Give the agent a clear name such as Support Assistant. The name is shown in your dashboard and analytics, and you can rename it later.
Add a website URL (optional)
Enter the address of the site you want the agent to learn from, for example https://www.your-company.com. Ultimo Bots checks that the URL is valid and reachable before it starts, so a wrong or private address is caught right away.
Upload documents (optional)
Drag files into the upload area or click Select Files. Supported types are .pdf, .txt, .doc, .docx and .rtf, up to 20 MB per file and 100 MB in total.

The New Agent dialog with a name and website URL filled in and a document upload area — The New Agent dialog, with a name and website URL filled in.

Website and documents work together. If you provide both, the agent learns from your site and your files in the same build. You do not have to choose one.

Create the agent

Click Create Agent. If you added a website, Ultimo Bots first confirms the site is reachable (you will see a short spinner). If you left both the website and documents empty, it asks Continue without knowledge? so you do not create an empty agent by accident.

Once the check passes, the dialog closes and your new agent appears in the list with a progress indicator. The build runs in the background, so you can keep working while it finishes.

What happens during the build

Building an agent is fully automatic. Depending on what you provided, Ultimo Bots does some or all of the following:

Reads your website. It looks for your sitemap first and otherwise follows the links it finds, reading up to around 100 pages.
Reads your documents. Uploaded files are parsed and added to the same knowledge base.
Sets up the assistant. It writes a starting persona, a welcome message, and two suggested questions, and picks a widget color based on your site.

Progress is shown live on the agent card. Typical builds finish in a few seconds to about a minute, depending on how much content there is.

While a build runs and after it finishes, the agent can be in one of these states:

State	What you see and what it means
Building	The card shows a progress percentage and a Building agent... label while the agent reads your knowledge. Progress updates live.
Ready	The build finished and the agent simply becomes a normal card in your list. It is ready to test, customize, and add to your site.
No content found	The agent was created, but no readable content was found at the URL or in the files. A separate onboarding message tells you so, and you can add a different source.
Build failed	The card shows a Build failed label. Something went wrong during the build. Adjust the source and try again.

After your agent is ready

Once the build finishes and the agent is ready, open it and keep going:

Test itChat with your agent to see how it answers before you publish.Add more knowledgeAdd more pages, documents, or Q&A so it can answer more questions.Customize the designMatch colors, welcome message, and starter questions to your brand.Add it to your siteInstall the agent on your website so visitors can start chatting.

Troubleshooting

The URL was rejected. The site must be public and reachable. Check for typos, remove any login wall, and make sure the page loads in a normal browser, then try again. No knowledge is added until the check passes.

No content was found. This usually means the page is behind a login, blocks crawlers, or has little readable text. Some sites that build most of their content with JavaScript can also yield little text for the agent to read. Point the agent at a different URL, upload documents, or add pages from the knowledge base instead.

The build failed. Builds are not retried automatically. Review the source you provided, then create the agent again or add the source from the knowledge base.