Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.aitextura.com/llms.txt

Use this file to discover all available pages before exploring further.

Knowledge is the foundation of accurate answers from your AI employee. Without a knowledge base, the agent relies only on instructions from the Persona and the model’s general knowledge. With a knowledge base, it provides fact-backed data: prices, terms, descriptions, specifications.

Persona

Define how your agent communicates

FAQ

Prepared answers and self-learning

Talents

Connect integrations and services

AI Employees

Manage all your virtual agents
Knowledge is static. After uploading, content does not update automatically. If the original document has changed, you need to delete the old source and upload the new one.
Persona defines how the employee responds (tone, style, behavior). Knowledge Base defines what it knows (facts, data, documents). These two tools work together: an instruction in the Persona can direct the employee to specific data from the knowledge base.

​
Two Places to Work with Knowledge

AI Textura has two related but distinct sections for working with knowledge:
SectionLocationPurpose
KnowledgeLeft sidebar menuCentral repository of all sources across all employees
Knowledge BaseTab in a specific employee’s settingsSelect which sources from the repository to connect to the agent
You can upload files from both places, but there is an important difference:
  • Uploading via the agent’s “Knowledge Base” tab — the file automatically lands in that agent’s subfolder inside Agents and is immediately connected to the employee.
  • Uploading via the “Knowledge” section in the left menu — the file goes to the folder you selected, but is not automatically connected to any employee. You must manually check the checkbox in the desired agent’s Knowledge Base tab.

​
”Knowledge” Section — Central Repository

The Knowledge section in the left sidebar is a shared repository of files and information sources across all employees. Here you can:
  • See the full picture: which sources are uploaded and which employees they are connected to
  • Organize information into folders
  • Add new sources
  • Delete outdated data

​
Interface

The screen is split into two parts: Left panel — folder tree. A hierarchical structure of all knowledge. At the top level is the root folder All knowledge. Click any folder to display its contents in the right panel. Right panel — file table. When you select a folder, a Knowledge files table appears:
ColumnWhat it shows
NameFile name, URL, or text source name
TypeElement type: Folder, Text, File, Web, etc.
UpdatedDate of last update
Above the table you will find:
  • Breadcrumbs — navigation through folder levels
  • Agent filter (All agents) — show sources connected to a specific employee
  • Add button — create folders and upload sources
Each item in the table has an actions button (three dots): rename, move, or delete.

​
Folder Structure

By default, the system organizes knowledge as follows: 
All knowledge
├── Agents
│   ├── [Employee 1 name]
│   │   ├── file.pdf
│   │   ├── text
│   │   └── ...
│   ├── [Employee 2 name]
│   │   └── ...
│   └── ...
├── FAQ
│   └── ...
└── [Your folders]
    └── ...
  • Agents — a system folder. Subfolders are automatically created for each employee when you add sources via the Knowledge Base tab in the agent’s settings.
  • FAQ — a folder for entries from the self-learning section that have been sent to the knowledge base.
  • Custom folders — create them at any level to organize by topic, department, or project.
Folder structure does not affect how the employee works — it searches across all connected sources regardless of their location. Folders are purely for your convenience.

​
Creating a Folder

1

Go to the Knowledge section

Open the Knowledge section in the left sidebar.
2

Click the create button

In the left panel, click the ”+” icon next to the Folders heading.
3

Enter a name

Provide a clear name — for example, “Price lists”, “Policies”, “Room descriptions”. The folder will appear in the tree.
You can create folders within folders for deeper organization, for example: Price lists → Summer 2026, Price lists → Winter 2026.

​
Filtering by Agent

Above the table there is an All agents dropdown. Select a specific employee to see only the sources connected to them. This is useful when you need to understand which knowledge a particular agent uses.

​
”Knowledge Base” Tab — Per-Employee Settings

Open the desired AI employee → Knowledge Base tab. Here you select which sources from the shared repository to connect to this agent.

​
Interface

The screen is split into two parts: Left panel — folder tree. All folders with knowledge sources. Each folder has a checkbox:
Checkbox stateMeaning
CheckedSources from this folder are connected to the current employee
UncheckedSources are not in use
Partially checkedSome sources inside are connected, some are not
Simply check or uncheck a box to connect or disconnect an entire folder. Right panel — file table. When you select a folder, it shows the Knowledge files table:
ColumnWhat it shows
AccessCheckbox — whether the source is connected to this employee
NameFile name, URL, or text source name
TypeSource type (Text, Folder, etc.)
UpdatedDate of last update
Above the table — breadcrumbs for folder navigation.

​
Connecting Sources to an Employee

1

Open the Knowledge Base tab

Go to the desired employee’s settings → Knowledge Base tab.
2

Find the relevant sources

In the left panel, locate the folder with the sources you need.
3

Check the box

Mark the checkbox next to the folder or individual file.
4

Save

Click Save. The same source can be connected to multiple employees.
Organize sources into folders — for example, one folder per employee or per topic. This makes access management easier as the number of sources grows.

​
Four Ways to Add a Knowledge Source

While in the desired folder (in either section), click the Add source button. Four options are available:

​
1. Upload Files

Upload documents from your computer. Supported formats:
FormatDescription
PDFDocuments, price lists, catalogs
DOCXWord text documents
TXTPlain text files
CSVData tables
When to use: you have ready-made documents — price lists, policies, service descriptions, manuals.
For best results, use structured documents with headings and lists. AI works much better with well-organized text than with a continuous stream of characters.

​
2. Text

Manual entry or paste text directly into the platform.
FieldDescription
Name (optional)A clear name for the source
ContentThe text itself — paste or type manually
When to use: you need to quickly add brief information — an address, contacts, business hours, special terms.

​
3. YouTube Video

Import a video transcript from YouTube. Enter the URL — the platform will automatically extract text from subtitles. When to use: useful information is contained in video reviews, presentations, or educational videos.
Import only works with videos that have subtitles — either automatic or uploaded by the author.

​
4. Web Page

Import content by URL. Supports regular websites, Google Docs, and Google Sheets.
ParameterDescription
URLPage address
Parse modeSingle page — only the specified URL. Full site — recursive crawl of all pages
When to use: the information is already published on a website or in a Google document.
To import from Google Docs or Google Sheets, simply paste the link — the system will detect the format automatically. For private documents, you will need to grant access to the platform’s service account — the address is shown in the import dialog.

​
Updating Knowledge

Uploaded sources do not update automatically. If a price list has changed, the employee will continue using the old data.
1

Find the outdated source

Go to the Knowledge section and locate the file.
2

Delete the old version

Using the actions menu (three dots), delete the outdated source.
3

Upload the new version

Upload the new file or re-import the page.
When you delete a source, it disappears from all employees it was connected to. If you only need to remove it from one agent, uncheck the box in that employee’s Knowledge Base tab instead of deleting the file.

​
Actions Summary Table

ActionWhere to do it
Upload a new sourceKnowledge (left menu) or Knowledge Base (agent tab)
Connect a source to an employeeKnowledge Base (agent tab) — via checkboxes
View all sources across all employeesKnowledge (left menu)
Delete a source from the systemKnowledge (left menu)
Disconnect a source from an employee (without deleting)Knowledge Base (agent tab) — uncheck the box

​
Usage Examples

A hotel uploaded sources to the Knowledge section and organized them into folders:
All knowledge
├── Agents
│   ├── Concierge (Telegram)
│   └── Support (website widget)
├── General documents
│   ├── House rules.pdf
│   ├── Price list 2026.docx
│   └── Room descriptions.txt
├── Restaurants and spa
│   ├── Restaurant menu.pdf
│   └── Spa services.txt
└── Transfers
    └── Routes and prices.txt
“General documents” and “Restaurants and spa” are connected to both employees. “Transfers” is connected only to the Concierge. When spa prices changed, the manager deleted the old file and uploaded a new one. Both employees immediately received up-to-date data.
Two agents: “Product Consultant” and “Customer Support”.
  • Catalog — product descriptions, specs, availability. Connected only to the Consultant.
  • Policies — returns, shipping, warranty. Connected only to Customer Support.
  • General — contacts, business hours, promotions. Connected to both.
When a new product was added, the file went into “Catalog”. The Consultant can immediately talk about it, while Customer Support does not even see it.
The school uploaded a class schedule as a CSV and connected it to the curator agent. Students ask “When is the next marketing class?” — the curator finds the answer in the table. The schedule is updated monthly: the admin deletes the old CSV and uploads the new one.

​
Best Practices

  • Clear file names. “Price list 2026 March.pdf” is immediately understandable. “Document_final_v3(2).pdf” is not.
  • Don’t upload everything. Each extra source is context the employee searches through. More irrelevant documents mean a higher chance of inaccurate answers.
  • Structure your documents. Use headings, lists, and tables. A long unstructured document is better split into several thematic files.
  • Test after uploading. Open a test chat and ask questions about the uploaded information.
  • Keep data current. Outdated data means incorrect answers. The employee does not know the information is stale.

​
FAQ

There is no limit on the number of sources. However, more sources means more tokens spent on search — only add what the employee truly needs.
No. All knowledge is static. If the original document has changed, you need to delete the old source and upload a new version. There is no automatic synchronization.
Yes. A source is stored in the Knowledge section and can be connected to any number of employees via checkboxes in each agent’s Knowledge Base tab.
It disappears from all employees it was connected to. If you only need to remove a source from one agent, uncheck the box in the Knowledge Base tab instead of deleting the file.
Check three things: (1) the source is connected — the checkbox is checked in the Knowledge Base tab; (2) the information is well-structured — use headings and lists; (3) the Persona includes an instruction like “answer based on knowledge base data, do not make things up.”
No. Folders are a tool for your convenience. The employee searches across all connected sources regardless of their folder location.
If the information already exists in a document, upload the file. If you need a brief reference (address, contacts, a short rule), use text. For current information from a website, import the web page.