Intelligent Assistant

The built-in AI assistant of the Citeck platform automates business processes, simplifies document management, and helps manage data.

Note

The Assistant is available in the Citeck Enterprise version under the following conditions:

  1. License: presence of the ai flag in the license

  2. User rights: the user must be in the GROUP_ai-feature-allowed group or have administrator/system rights

  3. Microservice: the ecos-ai microservice is connected and configured (see configuration)

User Interface

Main UI Features

  • Two operating modes: switching between universal and contextual assistant

  • Autocomplete with @: quick addition of records to the chat context

  • Context tags: visual representation of the added context

Hotkeys

  • Cmd+I (Mac) / Alt+I (Windows/Linux) - open/close AI assistant

  • Enter - send message

  • Shift+Enter - new line

  • Escape - close chat

  • @ - invoke autocomplete to add context

Settings

See configuration details

Universal Assistant

Universal AI assistant with automatic intent detection and context-dependent tool selection.

Supported tasks:
  • Artifact Assistant - creation, editing, and deployment of supported artifacts.

  • Business application generation - creating full-featured applications with data types and processes from a text description.

  • Script Writing Assistant - creating and editing scripts for various Citeck contexts.

  • Document analysis - information extraction and answering questions about documents.

  • Document comparison - analysis of differences between document versions.

  • Email writing assistant - composing business letters for leads, clients, and partners.

  • Customer 360 - comprehensive analysis of customers, deals, orders, and relationships.

Artifact Assistant

Data Types

  • Generation of data types based on text description.

  • Editing an existing data type. To do this, navigate to the data type card and add @record to the chat context.

  • Deployment of the data type in the system.

  • Questions and analysis regarding attributes, structure, and functionality of data types. For example: Tell me about the data type @record? What attributes and logic does it have?

Requests:

  • Create a data type for approving a vacation request with the manager

  • Add a “number of vacation days” field, let it be calculated automatically

  • What attributes are in the data type @record? How is the number of vacation days attribute calculated?

Business Application Generation

Creating full-featured business applications from a text description with automatic generation of all necessary components.

  • Application generation - creating full-featured business applications from a text description

  • Data types - automatic creation of data types with attributes, roles, and statuses

  • BPMN processes - generation of business processes for logic automation

  • Related entities - support for multiple related data types in a single application

  • Clarifying questions - interactive dialogue for collecting missing information

  • Deployment - automatic deployment of created artifacts in the system

Workflow:

  1. Describe business requirements in natural language

  2. AI will analyze the request and ask clarifying questions if necessary

  3. The system will suggest an application structure with data types and processes

  4. Confirm or request changes

  5. AI will automatically create and deploy all components in the system

Requests:

  • Create an application for managing vacation requests

  • I need a conference room booking system

  • Create an application for approving business trips with the manager and accounting

  • Create an equipment tracking system with reservation capability

Form Generation

  • New artifact type: the AI assistant can now generate and edit Citeck forms based on a text description

  • Dialog mode: Form generation with draft preview, the ability to refine and confirm before deployment

  • Validation: Automatic verification of the form structure before deployment

  • Business application integration: Form generation as part of the business application generation pipeline

Smart Context and Agent Mode for Artifacts

Agent Mode (Plan-and-Execute)

  • Complex requests: the AI assistant automatically determines the complexity of the request and switches to agent mode for multi-artifact tasks

  • Planning: Automatic generation of a plan with the ability to review and approve before execution

  • DAG execution: Step-by-step plan execution considering dependencies between steps (data type generation, forms, BPMN, search, analysis, validation, deployment)

  • User control (HITL): Plan approval, modification, skipping, and repeating steps via action buttons

  • Error recovery: Automatic retry with feedback and manual control upon failures

Smart Record Context Connection

  • Automatic detection: The AI assistant automatically finds and connects relevant artifacts from the workspace based on the user’s request, without explicit @-mention

  • Confirmation on ambiguity: When confidence is low, the system prompts the user to confirm the artifact selection

Script Writing Assistant

Specialized assistant for creating and editing scripts in Citeck.

How to Use

  1. Open any script editor (BPMN Script Task, computed attribute, etc.)

  2. Click the AI assistant button in the lower right corner of the editor

  3. In the input field that appears, describe what the script should do, or select a quick action

  4. The assistant will generate code with a detailed explanation of the logic

Supported contexts:

  • BPMN Script Task — scripts within business processes

  • Gateway / Sequence Flow Condition — process routing conditions

  • Computed Attribute — computed attributes of data types

  • Computed Role — dynamic determination of role participants

  • UI Action Script — client-side action scripts

  • Journal Formatter — journal cell formatting

Capabilities:

  • Script generation based on text description

  • Editing and improving existing scripts

  • Quick actions: code explanation, error fixing, optimization

  • Generation of computed attributes and journal formatters

  • API validation - verification that only available APIs for the context are used

  • Context-dependent generation - consideration of data type metadata, attributes, roles, and statuses

Requests:

  • Write a script for a BPMN task that sends a notification to the manager

  • Create a computed attribute for calculating the order amount

  • Explain what this script does

  • Fix errors in the script

  • Optimize this script

Document Analysis and Comparison

  • Document analysis based on their attributes and content

  • Version comparison of documents, highlighting changes and legal aspects

  • Answering questions about documents

  • Ability to specify multiple documents of the same data type in the context

Requests (document analysis):

  • Analyze @record for compliance with Russian legislation

  • Briefly tell me what this contract is about?

  • Are there any clauses that violate the rights of the parties?

  • Suggest improvements

Requests (document comparison):

  • What is the difference between the attributes of contract @record and @contract No. 4? Create a summary table with the differences

Requests (version comparison):

  • What changes were made in the latest version of the contract @record?

  • Compare version 1.0 and 2.5 in @record. What were the differences?

Email Writing Assistant

Composing business letters for various types of recipients with personalization based on interaction history.

  • Composing business letters for any entity in the system - for leads, clients, and employees

  • Personalization of letters based on interaction history and context

  • Professional signature with current user data

  • Ready-to-send letters with the ability to edit and send from the system

Requests:

  • Help me write a letter to the lead @record

  • Compose a response to the client regarding our proposal

  • Need to write a business letter to a partner

  • Change the letter style to a more friendly one

  • Write a letter to inform the client about the status of their support ticket @record

Customer 360

Comprehensive analysis of clients and counterparties with a complete picture of interactions and business relationships.

  • Comprehensive client analysis - complete picture of interactions with counterparties

  • Deal analysis - statuses, amounts, success rates, development trends

  • Order overview - frequency, volumes, product preferences

  • Payment history - analysis of payment discipline and amounts

  • Relationship assessment - health of business relationships and growth opportunities

  • Support ticket analysis - SLA control and problem identification

Requests:

  • Tell me about the client @record

  • What deals do we have with this counterparty?

  • Show all orders for client @record

  • What payments were made by this client?

  • History of interactions with the client

  • Complete information about the client

  • Analyze the client’s support tickets, are there any SLA issues?

BPMN Assistant

Specialized assistant for working with business processes in the BPMN editor.

  • BPMN generation of processes from text description

  • BPMN editing using AI

  • Contextual help based on the current process and data type

  • AI Task with direct LLM calls - AI tasks within BPMN processes can make direct calls to an LLM via BpmnAiService

  • Question-and-answer mode - the BPMN assistant supports free-form user questions

Requests:

  • Create a vacation approval process with the manager based on the data type

When generating a BPMN process, the AI assistant considers the selected data type in the BPMN editor to specify roles, statuses, and the semantic load of process elements.

Video: BPMN AI Task

Video: Implementing Spam Check via BPMN AI Task

Contextual Work

The more detailed you formulate your question/generation requirements, the more accurate the response you will receive.

Adding Context

Use the @ symbol to add additional context to requests:

  • @record - add data of the currently open record

  • @document_name - find and add a specific document by name or number.

Note

The @ search is performed only among records of the same data type as the currently open document. For example, if a Contract is open — other contracts can be searched via @contract No.230.

Try creating a equipment purchase request process using the intelligent assistant.

Video: Example of Creating an Equipment Purchase Request

Text Editing

The text editing assistant has 2 functions:

  • editing any attribute of a document;

  • интеграция с редактором Lexical.

The assistant allows writing new articles, editing them, changing current text, improving task descriptions, improving headlines, and more for working with text.

Editing Document Text Attributes

For example, let’s improve a task headline (1).

Invoke the assistant (2) and ask it to generate a more understandable headline. The result will be provided as a diff-view (visual differences between two data sets) (3). When clicking the Apply Changes (4) button, an update request will be sent to change the headline.

../_images/title_01.png

Integration with the Lexical Editor

The assistant can be invoked from the Lexical editor. The editor is used, for example, in the “Properties”, “Comments”, “Activities” widgets, when editing publications, and knowledge base articles.

../_images/lexical_01.png

In the editor, you need to invoke the assistant and ask it, for example, to change the article’s style, add or change something:

../_images/lexical_02.png

The result will also be provided as a diff-view (1), which shows what specifically was changed. In this mode, when clicking the Apply Changes (2) button, the text is immediately inserted into the current editor so it can be further edited, modified, and then saved.

Targeted Text Editing

You can select a paragraph and invoke the assistant specifically for that paragraph:

../_images/point_01.png

In this case, only the selected text will be changed. For example, let’s ask to add emojis to the text we selected:

../_images/point_02.png

The result will also be provided as a diff-view (1), which shows what specifically was changed, then apply the changes (2) and save the article.

../_images/point_03.png

Knowledge Base Search (RAG)

  • New context source - the AI assistant finds relevant information in connected knowledge repositories using vector search

  • ragSearch tool - search for relevant documents with filtering by Citeck types and repositories (includeRepoIds / excludeRepoIds)

  • ragGetDocument tool - retrieval of the full document content by identifier

  • sourceId protection - full content retrieval is only allowed for documents found via a preceding search within the same request

Citeck Documentation Search

  • Dedicated source - a separate search in the Citeck documentation repository, isolated from the general knowledge base search

  • SEARCH_DOCUMENTATION agent step - a new plan step for requests that require referring to official documentation

  • Content cleanup - automatic processing of documentation content before passing it to the LLM