Skip to content

Initial Setup

This guide walks through every step of the initial setup so your Clienta.ai workspace is fully configured before you start serving customers.

  1. Create an Account

    Go to app.clienta.ai/register. You can register with email and password or use Google OAuth for instant sign-in. If you choose email registration, check your inbox for a verification email and click the confirmation link.

    Registration Page

  2. Create an Organization

    On first login you will be prompted to create an organization. Enter your organization name and click Create. This is your isolated workspace — all team members, knowledge, and settings live here.

  3. Invite Team Members

    Go to Settings → Team and enter each member’s email address along with their role.

RolePermissions
AdminFull access — manage settings, billing, team, channels, and knowledge.
AgentHandle conversations in the inbox, view knowledge, and use the Test Bot.
MemberRead-only access to conversations and reports.

Go to the Knowledge page and click Upload Document. The system automatically processes each file into chunks and indexes them for semantic search (RAG).

FormatNotes
PDFText-based PDFs only. Scanned images require OCR pre-processing.
DOCXMicrosoft Word documents.
TXTPlain text files.

Maximum file size is 10 MB per document across all plans.

PlanMax Documents
Free3
Starter10
Growth100
Plus200
Pro500
Scale2,000
Enterprise5,000

Go to Settings → AI to configure how your bot responds.

Guardrail Mode controls how strictly the bot stays within your uploaded knowledge.

ModeBehavior
StrictAnswers only from uploaded documents. Declines questions outside the knowledge base.
HybridPrefers uploaded documents but uses the model’s general knowledge when no relevant document is found.
OpenAnswers freely using both documents and general model knowledge.

When Smart Model Routing is enabled, Clienta.ai automatically selects the most cost-efficient model for each query — using a faster, lighter model for simple questions and a more capable model for complex ones. This reduces cost without sacrificing answer quality.