AI AGENTS
Agents that know your business. A team ready for the next step.
Bring business knowledge, conversation support and enabled actions into one working experience. Configure the agent’s role and tone, support your team with useful suggestions, and keep human follow-up within reach.
Discuss your AI use caseConnect approved business knowledge with clear replies and a handover to your team.
AGENT WORKSPACE
Customer care assistant
Role · Knowledge · Response style
INSTRUCTIONS
Explain our services clearly. Ask for the missing detail before suggesting a next step.
Needs a person?
Keep the conversation moving with the right team.
CONVERSATIONAL AGENTS
Give recurring questions a capable first response.
Connect a conversational agent to supported messaging channels. Shape its purpose, approved answers and handoff instructions around enquiries, service guidance and the information your team needs next.
WhatsApp, web chat, Instagram and Messenger each retain their channel requirements. Confirm the enabled connection and test the intended conversation before launch.
SERVICE ENQUIRY
Can you help me choose?
AI assistant
Of course. How many people will use the service, and which channels do you need?
Purpose
Understand the need
Next step
Relevant answer or team handoff
Incoming call
Welcome. How can we help?
Listen
Identify the request
Respond
Use the configured scenario
Continue
Route to the right team
VOICE AGENTS
A voice journey with a clear destination.
Design supported voice scenarios around a greeting, the caller’s request, required details and the next response. Connect the voice agent to the enabled telephony service, with an agreed route to your team when human follow-up is needed.
Voice-agent activation, Arabic and English voices, Saudi voice options and conversation depth are confirmed separately with the connected voice service.
KNOWLEDGE DESK
Files your team relies on
PDF, DOCX, CSV and XLSX knowledge sources.
Pages that explain your services
Index relevant URLs and link the right knowledge.
Answers you have approved
Create focused question-and-answer pairs.
START WITH YOUR BUSINESS
A useful answer starts with useful knowledge.
Build the agent’s knowledge from approved files, indexed URLs and question-and-answer pairs. Keep sources relevant to the task and review answers against the material your business has approved.
Persona and instructions
Set the tone, priorities and boundaries that fit your customer experience.
Testing and refinement
Review responses against your source material and refine the instructions.
SPEECH AND TEXT
Write what the caller hears. Read what was said.
Text-to-speech turns approved scripts into voice prompts. Speech-to-text makes supported recordings and voicemail easier to review. These services connect voice experiences with the information your team can use afterwards.
TEXT → SPEECH
Your call matters to us.
Voice prompts · IVR · Supported campaigns
SPEECH → TEXT
00:08 · I would like to follow up on my enquiry.
00:16 · Please connect me to customer care.
Recordings · Voicemail · Reviewable text
IVR generation follows the PBX integration. Transcription and languages depend on activated services and recording quality.
ASSISTANCE IN CONTEXT
From a customer question to a considered reply.
Use the recent conversation to prepare a relevant reply. Agent Assist can surface response suggestions and available knowledge guidance; the employee reviews, edits or discards them before responding.
CUSTOMER QUESTION
Can we start with a small team?
We need business calls and WhatsApp, then more channels later.
SUGGESTED RESPONSE
Let us confirm your team size and the channels you need first. We can then review the relevant subscription, channel and usage costs together.
SENTIMENT AND SUMMARIES
Read the situation before choosing the reply.
Use available sentiment analysis to distinguish positive, neutral and negative messages. A concise summary and suggested replies help the employee find the issue and prepare a considerate next step.
Review sentiment alongside the conversation and any customer feedback. It is an inferred signal, not a customer satisfaction rating or a final assessment of the employee.
REVIEW BEFORE RESPONDING
I have asked twice and still have no clear update.
Negative sentiment · inferred
What matters now
Repeated follow-up. The customer needs a clear status update.
I understand. Let me review the request and clarify the next step.
Your team reviews suggestions before use.
AI AND WORKFLOWS
Connect the response to the next step.
Connect enabled AI steps with the customer record, conversation labels, tickets and follow-up workflows. Decide which information enters the step and which configured action follows: assignment, record updates or a supported message.
Keep ownership clear
A request can need an answer, more information or a handoff. Design each path around the team responsible for it.
01 · CONTEXT
A customer needs follow-up
02 · AI STEP
Summarise the need
03 · NEXT ACTION
Assign the configured follow-up
Workflow actions follow configured permissions and integrations.
AGENT INSTRUCTIONS
Source
Approved business knowledge
Missing answer
Ask or request team support
Commitments
Follow the approved limits
Human follow-up
The team takes it from here.
Keep the request, conversation and assigned owner available for follow-up.
BOUNDARIES AND HANDOFF
Give the agent a role. Keep the responsibility clear.
Set instructions for scope, missing information and escalation. Define what the agent may answer, which actions are enabled, and when the conversation should return to the team. Review autonomous operation against the permissions and scenarios you have approved.
AI ANALYTICS
See what happened. Find what needs attention.
Review available conversation logs, keyword frequency and sentiment distribution. For enabled voice analysis, inspect call summaries and transcripts after the call. Compare available agent indicators and customer feedback to identify where the team needs support.
Sessions
Daily activity and selected period
Resolution & handoff
How conversations conclude
Customer feedback
Available CSAT responses
Agent performance
Compare enabled agent indicators
TOPICS
Keywords and recurring topics
See which enquiries return and where clearer knowledge would help.
REVIEW
Sentiment and interaction records
Read sentiment alongside the conversation, summary and available transcript.
IMPROVEMENT
Team review and service quality
Use available performance indicators and CSAT to guide coaching and knowledge updates.
Voice and chat analysis, transcripts and reporting depend on the services activated for your workspace.
PERSONALITY AND OPERATION
Tune the agent to the job, not just the channel.
Configure its name, role and system instructions. Adjust tone, creativity and response priority to suit the use case, then manage deployment status. Keep service activation, access and usage review within your workspace’s enabled controls.
Role
Customer service
Tone
Clear and welcoming
Response priority
Balanced
Creativity
Aligned with approved answers
Review → Test → Activate
MODEL CONNECTIONS
Choose the connection that fits the job.
Select from the model connections enabled for your workspace. Confirm the supported provider, credentials and usage before activation; additional providers and custom integrations need their own availability review.
TEST BEFORE YOU EXPAND
A good launch begins with real questions.
Test common enquiries, unclear requests and cases that require a person. Review the response against its source, check the handoff and follow the available session and performance records. Improve the configuration before widening the deployment.
A known question
Is the answer grounded in the source?
An unclear request
Does the agent ask for the right detail?
A human is needed
Does the conversation reach the right team?
A test panel or configured workflow helps validate behaviour. Knowledge updates do not mean the underlying model trains itself automatically.
YOUR QUESTIONS
Before you activate your agent.
What is the difference between an agent and Agent Assist?
An enabled conversational agent responds within its configured scope. Agent Assist prepares suggestions for an employee to review, edit or discard before responding.
Can it use our own documents?
Build the knowledge workspace with approved files, URLs and question-and-answer pairs.
Does it work in Arabic and English?
Language support varies by the connected model and by chat, speech and analytics services. Confirm Arabic, English and any Saudi voice requirements for each module during setup.
Are voice agents and IVR the same service?
A voice agent follows the supported conversational scenario; IVR uses the configured telephony prompts and routing. Voice engines, speech services and PBX integration are confirmed separately.
Can it execute business actions?
Only configured and enabled actions are available. Confirm permissions, the target record or integration, and the next owner; a suggested action is not proof that it has been executed.
How do we organise an AI agent’s knowledge and handover?
Choose the knowledge sources, channels and handover rules around the conversations your team handles.
Start with one meaningful use case.
Tell us what customers ask, how your team responds and where assistance would help. We will review the relevant agent, knowledge and activation requirements with you.
Talk to the UCaaS teamAI services, provider usage and connected channels may have separate charges.
%20(3).png)