Quicker answers for customers and the public, with a person always close by.
We begin with the staff who answer: suggested replies, case details and summaries for the contact centre. Self-service comes later and only where it is safe, with firm limits and a handover that passes the whole conversation to a person.
For service teams in Malaysia that answer customers or the public on several channels and across Malay and English.
A contact comes in on any channel. The system recognises the request, the language and any sensitive subject, and looks up the case. Sensitive or out-of-scope contacts move to a specialist with the full history. In all other cases the assistant drafts a reply and suggests next steps, which a service officer edits and sends. Every contact is logged in the CRM, and a sample is checked for quality.
Before and after
How the working day changes
Today
With the system
Today
Today: Searching while the caller holds
Officers move between the CRM, the case system, the knowledge base and old emails to answer one question.
With the system
With the system: Context before the greeting
Case status, recent contacts and the relevant articles appear on screen before the officer answers.
Today
Today: Wrap-up takes over the shift
Each contact closes with minutes spent typing a summary and choosing codes, often from memory.
With the system
With the system: Drafts, never decisions
The assistant prepares the reply and the summary; the officer edits, sends and remains responsible.
Today
Today: Chatbots with no way out
Scripted bots go round in circles on questions they cannot handle, and callers reach a person already annoyed and telling their story again.
With the system
With the system: Always a way to a person
Self-service passes the conversation and the recognised request to a person as soon as it reaches its limits or the customer asks.
Today
Today: Queues split by language
Malay and English contacts wait in separate queues, so waiting times depend on who is rostered that day.
With the system
With the system: Malay and English by design
Replies are drafted in the language the customer used, in plain wording, and checked by the officer before they go out.
Workflow
How the flow runs
Assisted replies come first: the assistant prepares and an officer sends. Self-service follows later for well-defined questions, with the same controls and the same handover path.
Illustrative workflow
7 steps · 2 human checkpoints
Legend
System
AI agent
Person
Human checkpoint
Exception path
Service assistance: illustrative flow with a handover path
Read the diagram as text
The main path starts with a contact arriving on any channel, continues with reading the request and drafting a reply, and reaches the officer who checks and sends it. The contact is then logged in the CRM, and a sample goes to quality checks.
A person reviews every draft and approves any suggested action before it is sent: this is the human checkpoint on the main path.
Contacts flagged as sensitive, complex or outside scope leave the main path once the request is read, and go to a specialist with the full context. They join the main path again when the contact is logged.
01
Step 1: Contact arrivesSystem
Calls (with live transcription), chat, email, portal messages and scanned letters land in a single queue. Where a request needs it, the person is identified through the portal login or, for public services, MyDigital ID.
02
Step 2: Read the requestSystem
The system recognises the request, the language and the urgency, and flags subjects that must always reach a person (complaints, vulnerable customers, legal disputes). Case status is fetched through authenticated APIs.
03
Step 3: Hand to a specialistPerson
Sensitive, complex or out-of-scope contacts move to a specialist queue with the transcript, the recognised request and the case data attached, so nobody tells their story twice.
Exception path: Sensitive, complex or outside scope· Returns to step 6
Human checkpoint
The assistant gives no legal advice and takes no decision on entitlements. The service owner sets the handover rules, not the model.
04
Step 4: Draft a reply and next stepAI agent
The assistant prepares a reply from approved knowledge, in plain wording and in the language of the contact, and suggests actions it may prepare: booking an appointment, updating an address, sending a status update.
05
Step 5: Officer checks and sendsPerson
The officer edits the draft, accepts or rejects the suggested actions and sends the reply on the channel the customer used.
Human checkpoint
No reply leaves and no action runs until a person approves it. The officer’s edits are kept to improve future drafts.
06
Step 6: Log the contactSystem
A summary, the contact reason and the outcome are written to the CRM or case system, and the conversation is kept for the agreed retention period.
07
Step 7: Quality checksPerson
Team supervisors review a sample of conversations and drafts, note problems and update knowledge articles. Changes pass the evaluation set before release.
Components
Components we would deliver
01
Channel and telephony connections
Links to the contact centre platform, chat, email and portal, with live transcription of calls and one conversation history across channels.
Case data is fetched only once the person is authenticated, and only for that person. The drafting assistant works with a short list of actions, each with its own permission, and cannot change records itself.
Human review
In assisted mode, an officer approves every reply and action. In self-service, the handover triggers are explicit: sensitive subjects, low confidence, a question asked repeatedly, or a plain request to speak to someone.
The assistant gives no legal advice and decides nothing about entitlements, benefits or claims.
Auditability
Conversations, drafts, edits, and suggested and executed actions are logged with the model and prompt version, so a complaint can be traced step by step. The same logs show which conversations a personal data breach touched, so that the Personal Data Protection Commissioner can be notified within the 72 hours the Commissioner’s guidance sets.
Data protection
Where the Personal Data Protection Act 2010 (PDPA) applies, customers receive a notice in the national language and English, as section 7(3) requires, explaining how conversation data is used and kept. Transcripts and logs follow retention rules agreed with your data protection officer, personal data is masked in analytics, and nothing trains external models.
AI transparency
Where the service reaches people in the EU, they are told that they are dealing with an AI system, as Article 50 of the EU AI Act requires, and we show the same notice to every user. Anyone can ask for a person at any time, and replies that officers send remain your organisation’s own communication.
Measures
Measures we would agree
We agree the measures with you during discovery and compare them with the same queues before the assistant arrives.
Measures we would agree
Metric
Why it matters
How we would measure it
01Resolution at first contact
Why it mattersThe plainest sign that people got what they needed and did not have to come back.
How we would measure itRepeat contacts about the same case within an agreed period, from CRM data.
02Wrap-up time
Why it mattersSummaries and coding are where assistance saves the most time, and the conversation itself stays untouched.
How we would measure itTime spent after each contact, as recorded by the contact centre platform, per queue and channel.
03Handover quality
Why it mattersA handover only works if the specialist does not have to begin again.
How we would measure itSpecialists’ rating of the context they receive, and the share of handovers that need a call-back for missing details.
04Satisfaction
Why it mattersSpeed counts for little if people feel nobody listened.
How we would measure itSurveys after the contact and recurring complaint themes, compared by channel and language.
No targets are set before a baseline exists.
Rollout
Rollout in stages
Phase 01
Discovery
Contact reasons, channels, languages and systems. We choose frequent, low-risk requests and list the subjects that must always reach a person.
Exit criteria
Requests and handover rules agreed
Baseline taken for each queue
PDPA review with your data protection officer begun
Phase 02
Pilot of assisted replies
Context panels, drafts and summaries for one team on one or two channels. Officers send every reply.
Exit criteria
Drafts at the agreed quality
Officers trained and their feedback gathered
Logging and retention working
Phase 03
Self-service within limits
Chosen requests answered directly in chat or on the portal, with handover to the same team and the full context.
Exit criteria
Handover tested from start to finish
AI notice and PDPA notice live
Self-service results checked against complaints
Phase 04
Wider use
More requests, channels and languages, each passing the same evaluation gate.
Questions on applications, licences and assistance at federal agencies and local authorities, with MyDigital ID login for status checks and plain Malay and English replies.
Appointment, billing and administrative enquiries at private hospitals and clinics, with every medical question routed to qualified staff.
Illustrative example
Student enquiries at a private university
Situation
A private university could receive enquiries from applicants, students and parents by phone, chat, email and messaging apps, mostly as Malay and English messages. Peaks around intake and examination periods would leave officers typing summaries long after each contact.
System
The assistant would draft replies and summaries from one approved knowledge base inside the existing CRM, in the language the enquiry used. Self-service on the portal would cover application status, payment deadlines and document checklists.
Human control
Officers send every assisted reply. Appeals, complaints, welfare concerns and anything touching a student’s results always go to a member of staff, with the conversation attached.
What we would measure
Wrap-up time and resolution at first contact per channel, and specialist ratings of the handover context.
Malaysia
PDPA notice, consent and EU reach
A conversation with a customer is personal data under the Personal Data Protection Act 2010 (PDPA). The notice must be given in the national language and English, consent must be recorded where the Act requires it, and since 1 June 2025 a personal data breach is notified to the Personal Data Protection Commissioner, within 72 hours under the Commissioner’s guidance. We build the notice, the consent records and the conversation logs into the assistant from the first release.
The PDPA does not apply to the Federal and State Governments, so for their services we work to the government’s own policies and offer MyDigital ID login where a request needs it. Malaysian organisations serving consumers in the EU meet further rules: Article 50 of the EU AI Act on telling people they are dealing with an AI system, the European Accessibility Act for customer-facing digital services, applied through national laws since 28 June 2025, and the EU General Data Protection Regulation (GDPR) for transcripts of people in the Union.
No. We begin with assisted replies, so officers spend less time searching and typing and more time with the customer. Self-service is added only for well-defined questions, and a person is never more than one step away.
02How do you weigh self-service against handover?
We design the handover first. Every self-service request has explicit exit conditions and passes the full context to a person. We track the quality of handovers and repeat contacts, not only the number of conversations the assistant closes.
03Can it handle both Malay and English?
Yes, as a feature of the system. Drafts follow the language the customer used and are written in plain wording. Your officers review each draft before it is sent, and knowledge articles are kept in both languages.
04Must we tell people they are talking to an AI system?
If the service reaches people in the EU, Article 50 of the EU AI Act has required since 2 August 2026 that they are told when they interact directly with an AI system. We design the notice, the option to reach a person and the logging from the start, and we recommend showing the same notice to customers in Malaysia.
05Does it work with the CRM and contact centre platform we have?
That is the aim. Where possible we build inside your service console, for example as components in Salesforce Service Cloud, and connect your telephony and case systems through their APIs.
06Where should we begin?
With your contact reasons. Frequent, low-risk requests make a sound first scope for assisted replies. For our approach to agents, see AI Agents & Automation.
Start with your contact reasons
Share your most common contact reasons, your channels and your languages. We will suggest where assisted replies help first and where an officer should remain the first point of contact.