Overview
Qlar is an AI agent platform that lets you build, train, and deploy intelligent assistants, with no coding required. Whether you're building an internal knowledge assistant for your employees, an analytics agent that answers questions about your data, or a customer support agent, Qlar gives you all the tools in one place.
This guide will walk you through the core concepts so you can start building with confidence.
What is an Agent?
An agent is your AI assistant. It has a name, a personality, knowledge you've provided, and the ability to respond to users through one or more channels (like a web widget or WhatsApp).
Think of it like hiring a new team member: you define their role, give them the information they need, and choose how they interact with customers or colleagues.
In these guides you build Atlas, the operations assistant of a fictional company called Kirana Group. Atlas helps project managers, site supervisors, and procurement and finance staff. It tells them who has to approve a purchase, what a capex request needs, and how a site safety rule works, based on Kirana's own SOPs.
How Qlar Works
Qlar is built around a simple three-step loop:
1. Configure
Set up your agent's identity and behavior in the CMS dashboard. You define:
- Persona: the agent's name, tone, and personality
- Knowledge: documents, websites, and videos the agent can draw from (and, through plugins, your databases)
- Channels: where users will talk to the agent (web widget, WhatsApp, Instagram, etc.)
2. Train & Test
Chat with your agent in the preview, or use the Simulation tool to run test conversations before going live. See exactly how it responds, tweak its behavior, and make sure it knows what it needs to know. For Atlas, that means checking that it names the right approvers for a $2.5 million project before employees rely on it.
3. Deploy
Publish your agent and it's live instantly. You can monitor every conversation, track usage, and keep improving over time.
The CMS Dashboard: Your Control Center
When you open an agent in the CMS, its settings are grouped in the left sidebar. The groups follow the life of an agent: define who it is, give it knowledge and abilities, connect it to your users, test it, then measure how it does.
| Group | What it does |
|---|---|
| Home | Dashboard shows usage at a glance. Setup is a quick setup page with a live chat preview next to it. |
| Persona | Who the agent is: Profile, Behavior, Appearance, and About. |
| Conversation | How the agent talks: Introduction, Chat, Voice, Escalation, Personalization, and Model & Context. |
| Knowledge | What the agent knows: Documents, Videos, and Websites, plus Retrieval, Chunks, and Citations to tune how answers are found and cited. |
| Skills | Written procedures for specific situations (Skills) and the files they can use (Artifacts). |
| Tools | What the agent can call while answering: Plugins, Custom API, MCP, Other Agents, Web Search, and Web Scraping. |
| Channels | Where users reach the agent: Embed (a chat widget for your website), WhatsApp, Instagram, Agent-to-Agent, and Agent Keys. |
| Automations | Tasks the agent runs later on its own: Monitor and Policy. |
| Simulation | Test conversations before going live: Scenarios holds your test scenarios in folders, with their Results on a second tab. |
| Analytics | How the live agent is doing: Analysis, Feedback, and Logs. |
| Users | Who chats with, can access, or helps build the agent: Participants, Members, Contributor, Permissions, and Availability. |
Some items only appear when the feature is turned on for your workspace, and some (such as Members and Availability) are only visible to the agent owner. For a field-by-field reference of every screen, see Help.
Agent Types
When you create a new agent, you pick a type that sets up the right defaults for your use case. The list of types you see can differ per workspace; common ones include:
- SOP and internal knowledge: help employees find company information, policies, and procedures instantly. Atlas is this type.
- Data analytics: analyze data, generate insights, and answer business intelligence questions in plain language
- Customer support: answer customer queries, resolve issues, and provide instant 24/7 support across all channels
- Custom solution: build a tailored agent for your unique business needs and specific workflows
Each type comes pre-configured with relevant questions and sensible defaults, so you're not starting from zero.
What You Need to Get Started
- A Qlar account (sign up at qlar.ai)
- About 10 minutes for the first two guides. You do not need any documents for them.
- Later, content for your agent: for example your SOPs as PDF or text files, or the web pages your employees rely on
Next steps
Just signed up? Welcome to Qlar explains the page you see first, and Quick Start: Your First Agent explains the guided Create your first agent flow, which builds a sample data agent for you.
- Create Your First Agent: build Atlas with the New Agent wizard.
- Setup and Publish: give Atlas Kirana's brand and make it live.
- Configuring Your Agent: when you are ready to go further, follow the step-by-step journey from the basics to advanced features.