AI agents for customer engagement

Hire AI agents for collections, customer service and call quality.

They follow your rules, cover evenings and weekends, and hand over to your team when a person is needed. For financial institutions, enterprises and BPOs across Indonesia and Malaysia.

Singapore · Jakarta · Kuala Lumpur · Tokyo

Working with financial institutions, enterprises and BPOs in Indonesia and Malaysia

TA Asset BCA Finance Staffinc Ryt Bank Mekari Linksindo Cartenz
01 · The challenge

Customer engagement puts pressure on your team.

Reaching customers, covering support hours, reviewing conversations and planning collections all compete for the same people.

01

Low contact rates

Outdated numbers, unanswered calls and poor timing limit how many customers you reach.

02

Gaps in coverage

Customers need help in the evening and at weekends, often outside your team’s working hours.

03

Limited QA coverage

Supervisors review a small sample of calls and chats. Problems in the rest go unnoticed.

04

Unclear collection priorities

Without a clear way to prioritise accounts and plan follow-ups, collectors spend time on low-value work and miss recovery opportunities.

Turnover

Staff turnover adds recruitment and training costs, and new agents need time to learn your products and procedures.

02 · Solutions

Pick the job. We supply the agent.

AI agents for debt collection, payment reminders, customer service and call centre quality assurance. Each one is trained on your products and procedures, and your team sets its scope and reviews the results.

K
KenCollection agent

Follow up overdue accounts and agree how they’ll be paid.

Ken contacts customers, negotiates within your limits and follows up on payment commitments.

More about AI debt collection
OverdueNegotiationPromise-to-payRuns on SmartCollect
01

Prioritises accounts

Prioritises accounts by balance, age and the customer’s payment history.

02

Follows up overdue payments

Follows your contact schedule until an account is settled, restructured or passed to your team.

03

Negotiates within your rules

Agrees payment dates, instalments and waivers within your approved limits, and records the customer’s commitment.

04

Sends payment links

Sends a payment link during the call and records the outcome against the account.

Result

4× cumulative recovery against a human team on the same accounts. See the results →

H
HanaCustomer service agent

Remind customers before payments fall due.

Hana calls customers ahead of their due date, makes sure they know what to pay and when, and records what they say.

More about AI payment reminders
Upcoming paymentsRenewalsAppointmentsRuns on SmartRobo
01

Reminds before the due date

Calls customers about upcoming payments, renewals and appointments on the schedule you set.

02

Confirms the details

Checks the customer knows the amount and due date, and answers questions using your approved information.

03

Records the response

Logs whether the customer confirmed, asked for help or needs another call.

04

Passes on what needs a person

Customers with a problem or a dispute go to your team with a summary of the call.

Languages

English, Indonesian, Malay, Mandarin and Japanese, with switching during a call.

H
HanaCustomer service & tech support agent

Answer customers in and out of office hours.

Hana handles customer enquiries during and outside office hours, and makes outbound calls for verification, surveys and follow-ups.

More about AI customer service
Customer serviceTech supportHuman handoffRuns on SmartRobo
01

Customer service

Answers incoming calls about your products, services and policies using your approved information.

02

Tech support

Guides customers through troubleshooting steps, checks the result and explains what to try next.

03

Verification and surveys

Calls customers to verify details, confirm information and collect survey responses using your approved questions.

04

Hands over with context

Unresolved issues go to your team with a summary of the problem and the steps already tried.

Coverage

Evenings and weekends included, with 24-hour coverage where you need it.

M
MayaQA agent

Review every call and chat, not a sample.

Maya reviews calls and chats from your staff and AI agents against your QA criteria.

More about AI quality assurance
100% QAComplianceCoachingRuns on SmartQA
01

Scores calls and chats

Grades each conversation against criteria built from your own scorecards.

02

Flags compliance issues

Flags potential breaches of your rules for a supervisor to review.

03

Links scores to evidence

Each score points to the relevant part of the recording or chat, so reviewers can check it.

04

Provides coaching feedback

Gives supervisors specific feedback from calls and chats to discuss with each agent.

Start here

Maya can review your existing team’s calls and chats as a standalone service, including telesales calls and sales leads.

Hana, Ken and Maya are working names. You choose the names and voices for your deployment.

03 · Collections in depth

From overdue to paid, on your rules.

AI debt collection is where we go deepest. Ken works the accounts, and SmartCollect runs the campaigns around him.

01 · PRIORITISE

Pick the right accounts

Rank accounts by balance, age and the customer’s payment history.

02 · REACH

Call on your schedule

Rolling or batch campaigns, within your calling hours and attempt limits, with WhatsApp follow-up.

03 · AGREE

Negotiate within limits

Dates, instalments and waivers within your approved limits, with the promise recorded.

04 · COLLECT

Get paid

A payment link sent during the call. Payments post to the account and close the promise.

Campaign controls
  • Rolling and batch campaigns
  • Calling hours, attempt limits and retry intervals per client
  • Outcomes and retries tracked per account
  • Promises to pay tracked until they’re kept or broken
  • Voice and WhatsApp outreach
AI Collection Manager

Manage campaigns by asking.

Ask the AI Collection Manager to review performance and set up or adjust campaigns in plain language, so a small team can run a large book.

More about AI debt collection
04 · Results

Results from live client programmes.

Volume and recovery from collections programmes running on Kataliz, and how clients grow from their first use case.

Campaign volume
2.84M
calls in one quarter

Calls to 392,000 customers for one collections client, May to July 2026.

Recovery
4×
0.8% → 3.2%

A human team versus Kataliz on the same accounts, for a debt buyer in Jakarta, over the full placement period.

QA → Sales insight

A SaaS company

Started with quality assurance on telesales calls in April 2025, then added lead scoring across 5,689 sales chats.

QA → Collections

A BPO

Started with quality assurance on the helpdesks it runs for clients, then added outbound collection calls.

Collections → Service

A debt buyer

A client since October 2025, now adding inbound service calls and chats.

Call volume counts calls placed across one client’s campaigns, May–July 2026, not calls connected. Recovery compares one debt buyer’s accounts handled by a human team and by Kataliz, cumulative over the full placement period. Results depend on the portfolio.

05 · Day-to-day

Your team sets the rules and reviews the results.

Set the working hours, approved responses and escalation rules. Use call records and QA scores to review performance.

Flexible hours

Schedule support for evenings and weekends, with 24-hour coverage where required.

Your customers’ languages

English, Indonesian, Malay, Mandarin and Japanese, with switching during a call.

Contact and escalation rules

You set calling hours, attempt limits, approved offers and when a person should take over.

Call records

Review recordings, transcripts, scores and outcomes to check how every call was handled.

06 · Who it’s for

Built for teams with customer conversations to run.

Financial institutions

Banks, multifinance companies, lenders, collection agencies and debt buyers.

Where to start

Choose one portfolio or support queue. Agree the rules, escalation points and success measures before the pilot.

Enterprises, contact centres & BPOs

Teams can start with QA, then add customer support or collections as needed.

Example QA scorecard

Opening and identity · tone · privacy and disclosure · question handling · recorded commitments · outcome accuracy. Built from your rules.

07 · Next step

Plan your first pilot.

Start with one job. Review the results before deciding which agent to add next.

01

Pick the job

Debt collection, payment reminders, customer service or quality assurance.

02

Define the pilot

Agree the call volume, system access and success measures.

03

Book the kickoff

Confirm your project owner, access and rollout plan.

Timing depends on scope, systems access and testing.

08 · FAQ

Questions buyers ask us.

What does Kataliz AI do?

Kataliz AI provides AI agents that handle customer conversations: debt collection, payment reminders, customer service and tech support, and quality assurance of calls and chats. We set them up and run them with your team, for financial institutions, enterprises and BPOs in Indonesia and Malaysia.

Which languages do the AI agents speak?

English, Indonesian, Malay, Mandarin and Japanese. Indonesian and Malay are handled as separate languages, and an agent can switch language during a call.

Do the AI agents replace our team?

No. Your team sets the rules, such as calling hours, approved offers and when a person should take over, and reviews the results. When a customer needs a person, the agent passes the issue to your team with a summary.

How does AI debt collection work with Kataliz?

Ken, the collection agent, prioritises overdue accounts by balance, age and payment history, calls customers on your schedule, negotiates payment dates and instalments within your approved limits, records the promise to pay and sends a payment link during the call. On the same accounts of a debt buyer in Jakarta, cumulative recovery rose from 0.8% with a human team to 3.2% with Kataliz.

Can Kataliz review our existing call centre’s calls?

Yes. Maya, the QA agent, scores your team’s calls and chats against criteria built from your own scorecards, flags potential compliance issues for a supervisor to review, and gives coaching feedback for each agent. It can run as a standalone service.

Which industries does Kataliz work with?

Banks, multifinance companies, lenders, collection agencies and debt buyers, as well as enterprises, contact centres and BPOs. Clients include TA Asset, BCA Finance, Staffinc, Ryt Bank, Mekari, Linksindo and Cartenz.

How do we start, and how long does it take?

Start with a pilot on one job: pick the solution, agree the call volume, system access and success measures, then book the kickoff. Timing depends on scope, systems access and testing.

Where is Kataliz based?

Kataliz is headquartered in Singapore, with offices in Jakarta, Kuala Lumpur and Tokyo.

Put an AI agent on your hardest customer conversations.

Book a 30-minute call. We’ll look at your operation, show you an agent at work and agree what a pilot should prove.

Book a call