Documentation Index

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

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

Features

Prev Next

Document Number

Revision Number

Revision Date

KN. GU.18.EN

Rev14

03.04.2026


This document outlines the features available in AI Agents (NLU-Based) and AI Agents (Hybrid/LLM-Based) packages.

1. Deployment

Feature

Description

AI Agents (NLU-Based)

AI Agents (Hybrid/LLM-Based)

On-Cloud and On-Premises Availability

Deploy the platform in your preferred environment — choose cloud-hosted for rapid time-to-value and automatic scaling, or on-premises for complete data sovereignty and compliance requirements



2. Core AI Platform

Feature

Description

AI Agents (NLU-Based)

AI Agents (Hybrid/LLM-Based)

Multi AI Agents & Tools Orchestration Platform

Multi-agent orchestration with autonomous task planning and hierarchical agent delegation — each agent focuses on a specific task while multiple agents collaborate to accomplish complex goals. Seamlessly integrate external tools (CRM, ERP, APIs) and enable agent-to-agent handoff for end-to-end task automation

Knowledge Base (RAG)

Enterprise Retrieval-Augmented Generation with advanced RAG techniques — ingest corporate documents, FAQs, and policies with Intelligent Document Processing (IDP) for visually rich content (tables, forms, scanned PDFs) into a vector database. Leverage hybrid search (semantic + keyword) for precise retrieval. Deploy as a standalone Knowledge Base or enable AI Agents with the Knowledge Retrieval tool to generate accurate, grounded responses with minimized hallucination.

Input & Output Guardrails

Comprehensive AI safety controls for both customer inputs and model outputs — detect prompt injection, enforce content moderation, mask PII, ensure brand compliance, and block harmful outputs. Features on-the-fly PII masking with our 100% in-house built masking engine for complete data control and compliance.

AI Testing

Automated conversation testing with dual approaches: Agent-as-User to simulate realistic customer interactions and test agent responses against synthetic conversations, or test from real production conversations to validate with actual customer behavior. Catch regressions, measure accuracy, and ensure consistent agent performance and deploy with confidence.

Advanced Conversation Routing

Hybrid NLU + LLM routing — the best of both worlds. High-confidence matches execute instantly via NLU; ambiguous inputs get LLM-powered disambiguation with clarifying questions. Route complex intents to specialized AI Agents and hand back to NLU seamlessly when done. Reduces false fallbacks dramatically while protecting your NLU investments. Migrate from traditional VA to Agentic AI gradually, not painfully.

NLU-Based Conversational Flow Platform

Production-proven Natural Language Understanding platform for building deterministic, auditable conversation flows with precise intent recognition and entity extraction.

Advanced NLU Engine

State-of-the-art ML-powered NLU with high-accuracy intent classification, hierarchical entity extraction, spell correction, and contextual understanding. Feedback-driven learning from unmatched utterances continuously improves model accuracy.

Low-code Flow Designer

Intuitive visual drag-and-drop builder to design flexible conversation scenarios without coding. Create flows with nodes for questions, conditions, API integrations, and agent handover. Supports rich content outputs: quick replies, carousels, cards, location sharing, and dynamic buttons. Leverage reusable components (intents, entities, sub-flows) for faster development. Built-in simulator lets you test both chat and IVR experiences directly in the design screen before deployment.

Post Session Flow

Workflow automation triggered after conversations — auto-update CRM, create tickets, send surveys, trigger webhooks, and execute post-call processing without manual intervention

Multichannel Connectivity

Connect to customers across multiple channels like WhatsApp, Web Chat, Facebook, Telegram, Teams, SMS, IVR, and more from a single unified interface. Channel-agnostic design means build once, deploy everywhere.


3. Language & Text Intelligence

Feature

Description

AI Agents (NLU-Based)

AI Agents (Hybrid/LLM-Based)

Sentiment Analysis (Text)

Analyze conversation sentiment in real-time to understand customer satisfaction and emotional state

Language Detection (Text)

Automatically detect the language of incoming messages to route or respond appropriately

Realtime Translation

Translate conversations on-the-fly, enabling agents and AI to communicate with customers in their preferred language


4. Security & Authentication

Feature

Description

AI Agents (NLU-Based)

AI Agents (Hybrid/LLM-Based)

Manually Managed Email Accounts

Create and manage platform user accounts with email-based authentication

SSO (OpenID)

Enterprise Single Sign-On via OpenID Connect — integrate with Azure AD for seamless, secure access with existing corporate credentials. Simplify user management and enforce centralized authentication policies.

LDAP/LDAPS

Active Directory integration via secure LDAP/LDAPS — centralized user provisioning, group-based access control, and compliance with enterprise identity governance policies.

Data Masking

Flexible PII protection with two masking modes: Static masking permanently redacts sensitive data (credit cards, phone numbers, national IDs) in stored conversations for strict compliance. Dynamic masking preserves original data in database but masks on-the-fly when displayed in UI. User voice messages are also masked for voice channels.


5. Generative AI Features

Feature

Description

AI Agents (NLU-Based)

AI Agents (Hybrid/LLM-Based)

Utterance Generation

AI-powered training data augmentation — generate diverse utterance variations for intents in seconds, accelerating NLU development and improving model coverage.

LiveChat Summarization

Auto-generated conversation summaries during live agent handover — agents instantly get full context without reading entire chat history, reducing average handle time and improving resolution quality.

Reporting Summarization

On-demand conversation summaries — generate concise summaries of completed conversations anytime for CRM logging, post-call review, or quality assurance.

Response Suggestion

AI-powered response enhancement to generate alternative phrasings, improve tone, and multiply static response variations for a more natural, engaging conversation experience.

Generative Agent Design

Design conversational agents using natural language prompts instead of rigid flow structures and let AI automatically generate relevant intents and utterances.

Generative Coding Assistant

AI-powered code generation for flow development — generate JavaScript code blocks, debug issues, and build custom logic with natural language prompts directly within the flow designer.

Ask LLM

Direct access to LLM capabilities within conversation flows for dynamic, context-aware responses. Supports structured output (JSON) for seamless integration with downstream logic and API calls


6. Add-On Features

These features are available as paid add-ons.

Feature

Description

AI Agents (NLU-Based)

AI Agents (Hybrid/LLM-Based)

Data API Access

Automated daily exports of conversation data and audio recordings — hourly generated reports and audio files with API access for on-demand retrieval. Power your BI dashboards, feed custom analytics pipelines, and meet legal compliance requirements for conversation archiving.

Outbound Calling

Proactive voice outreach — initiate automated outbound calls for appointment reminders, payment notifications, surveys, and campaign calls. Scale your outreach without manual dialing.

Sestek-provided LLM Keys

Use Sestek-managed LLM API keys for generative AI features without bringing your own keys

Third-party TTS

Integrate premium text-to-speech providers (e.g., ElevenLabs, Azure) for enhanced voice quality

Private LLM / 3rd Party Engine Connector

Bring Your Own Model — connect private LLM deployments, fine-tuned models, or third-party AI engines for complete control over AI capabilities and data privacy.