Hamsa Agents · Chat

Arabic Chat Agents

The same Hamsa agent, over text, inside your own website or app. You build the chat interface with the Web SDK, and the SDK hands you the messages to show in it.

  • Text chat through the Web SDK
  • Your interface, no widget to adopt
  • Same agent as voice
  • Chats in Call History
  • Live monitoring of active chats

Capabilities

An Arabic AI chatbot built from the agent you already have

01

Text conversations

Chat is a channel next to phone and web voice. The session runs as text instead of audio, and the agent replies in writing.

02

Your interface, your app

There is no chat widget to adopt. You build the chat window with your own markup, styling and placement, and the SDK hands you the messages.

03

One agent for voice and chat

One flag in the SDK turns a voice session into a chat one. The agent, its knowledge and its tools stay the same, so there is no second bot to keep in step.

04

Answers from your documents

Agents answer from a knowledge base of your PDF, Word, text, HTML and EPUB files, switched on per agent.

05

Actions through your APIs

Function and MCP tools call your systems, so the agent can look up an order or book a slot. Credentials stay in encrypted project secrets.

06

Every chat on record

Chat sessions appear in Call History with their own channel filter, and active chats can be monitored live.

How it works

From a Hamsa agent to an enterprise Arabic chatbot in your app

What it takes to put an agent behind your own chat window, in order.

  1. 01

    Build the agent

    Write the agent’s instructions in Hamsa, then add its knowledge base and the tools it needs to reach your systems.

  2. 02

    Create a public key

    Make a public key for the environment your app runs against and limit it to your app’s domains. It is made for the browser: it can only start sessions.

  3. 03

    Install the Web SDK

    Add @hamsa-ai/voice-agents-sdk from npm or a CDN, and connect it to your agent by its ID.

  4. 04

    Start a chat session

    One flag starts the session as chat instead of voice. The SDK hands you each message for your interface.

  5. 05

    Render it your way

    Show the conversation in your own chat window, in your product’s design, wherever it belongs in the page.

  6. 06

    Monitor and review

    Follow active chats live, and find past ones in Call History with the chat channel filter.

At a glance

Chat agent settings at a glance

Chat runs through the Web SDK. The embeddable script widget and the public agent link are voice experiences.

SettingDetail
ChannelChat, alongside phone and web voice
InterfaceBuilt by you in your website or app; no hosted chat widget
SDK@hamsa-ai/voice-agents-sdk from npm or a CDN, one SDK for voice and chat
Knowledge basePDF, DOCX, DOC, TXT, HTML and EPUB files, switched on per agent
ToolsFunction tools for HTTP APIs and MCP tools, with credentials in encrypted project secrets
Language modelConfigurable per agent: OpenAI, Gemini, Groq, DeepMyst or any OpenAI-compatible endpoint
ReviewCall History with a chat channel filter; live monitoring of active chats
AccessProject roles (Owner, Admin, User); a public key in your app, a secret key on your server

Use cases

Where an Arabic customer service chatbot earns its place

01

Customer support

Answer questions from your knowledge base, check orders and accounts through your APIs, and guide customers through your site.

02

Sales assistant

Qualify leads in conversation, recommend products, and help customers through the buying journey.

03

Bookings

Help users schedule appointments, make reservations or change an existing booking through your booking system.

04

Internal assistants

Support employees with HR and IT questions, internal knowledge bases and internal workflows.

Example chat

An order question, answered in the app

An illustrative chat in a retailer’s app. The agent checks the order through a tool connected to the retailer’s system.

Example · order status chat

Example

callerالسلام عليكم، وين وصل طلبي؟ رقمه ٤٨٢١٣

tool: orders.lookup · order 48213 found

hamsa agentوعليكم السلام. طلبك رقم ٤٨٢١٣ خرج للتوصيل اليوم، ومتوقع يوصلك قبل الساعة ٦ مساءً.

callerممكن أغيّر العنوان لمكتبي؟ حي العليا، شارع التحلية

tool: orders.update_address · outcome: address updated

hamsa agentتم تحديث العنوان إلى حي العليا، شارع التحلية. هل تحتاج شيئًا آخر؟

FAQ

Questions about Arabic chat agents

Is there a ready-made chat widget?

+

No. You build the chat interface in your own website or app, and the Web SDK hands you the messages to display. The embeddable widget and public link that Hamsa provides are for voice conversations.

How is a chat agent different from a Hamsa voice agent?

+

It is the same agent on a different channel. One flag in the Web SDK starts a session as chat instead of voice, and the agent keeps its knowledge and tools. You do not maintain a separate chatbot.

Can an Arabic chat agent take actions in our systems?

+

Yes, through tools. Function tools call your HTTP APIs and MCP tools connect to MCP servers, so the agent can look up an account or update a record during the chat. Credentials are stored as encrypted project secrets and referenced by name.

Where do chat conversations show up?

+

In Call History, next to phone and web calls, with a channel filter for chat. Active chats can be monitored live by team members whose role allows live monitoring.

Which key goes in a browser app?

+

A public key. It is made to sit in a page: it can only start sessions, it works in one environment, and you can limit it to your domains and to the agents it serves. Secret keys stay on your server, and a request carrying one from a browser is refused.

Which language model does a chat agent use?

+

The model is configurable per agent. You can choose models from OpenAI, Gemini, Groq or DeepMyst, or point the agent at any OpenAI-compatible endpoint. Gemini is the default.

Keep reading

Partner ecosystem

Trusted Across the
Technology Ecosystem

Hamsa works with leading technology companies, integrators, and solution providers to bring enterprise Voice AI to organizations across the region.