- 10 hours ago
Would you trust AI to make decisions or spend your money?
As AI moves from advice to action, are Malaysians ready to hand over control? Cynthia Ng speaks to Dawn Lai, CEO of Experian Information Services Malaysia, about agentic commerce, consumer trust and how businesses must compete when AI agents influence purchasing decisions.
As AI moves from advice to action, are Malaysians ready to hand over control? Cynthia Ng speaks to Dawn Lai, CEO of Experian Information Services Malaysia, about agentic commerce, consumer trust and how businesses must compete when AI agents influence purchasing decisions.
Category
🗞
NewsTranscript
00:09Hello you are watching Awani Review my name is Sinti Ng. Now most of us are already using AI to
00:15get answers recommendations and even for advice but the next shift will be far bigger whereby
00:21instead of just giving us answers AI could ultimately increasingly act on our behalf
00:27that's the idea behind agentic AI and just how ready are we to hand over that kind of control to
00:33an AI agent so joining me today to talk more about this is Don Lai CEO of Experian Information Services
00:40Malaysia Experian has just released a new research looking at 13 markets including Malaysia and how
00:47consumers feel about agentic AI in commerce and what this shift could mean for consumers as well
00:55as businesses so thank you so much Don for coming on the show today great to have you here yeah
00:59glad
00:59to be here thank you for inviting me Cynthia so let's start with where we are today what could
01:04an AI agent realistically do for me now that wasn't possible or practical a few years ago yeah if you
01:11really look at it's right the the most of us nowadays rely a lot on AI and we are definitely
01:19leveraging
01:20AI's a lot in terms of using it for comparing product findings in terms of the financial terms
01:28understanding in terms of what are these things can do for us right in many years the AI that you
01:34have is a different types of AI even if you look at it the AI itself has gone through many
01:39rows of
01:39evolutions so the the current AI is a generative AI it does help us to make reason it help us
01:48to provide
01:49content helping us to do comparisons but the next evolutions is the ones that we talk about in the study
01:56how the agentic AI would change how the ways that we make purchases how the way we will actually shop
02:03in terms
02:04of commerce so we are not exactly at the point where let's say an AI can independently take out a
02:11loan on
02:11my behalf we are not there yet I would say that we are starting to see in some other market
02:17and based
02:18on the study that we have done if you look at the trust level on the AI is there definitely
02:25when come to
02:27comparing products getting advisors getting recommendations but then when you come to and
02:34trusting the agents acting on your behalf that trust level has come down to 50% because based on
02:41the study we've seen that the trust level is almost 7.6 out of 10 the trust score is very
02:47high but when you
02:49ask them that would you allow the agents give them the full autonomy to helps you to do something that
02:56very much depends what we have seen in the study is that low risk high convenient tasks people are
03:04still comfortable and trusting the agents to do it but even come to high risk financial transactions
03:11then the trust level will start to come down the car comfort level need to build especially as the
03:17trust level growth right so this picking the data that you just mentioned from this report the rise in
03:23agentic AI commerce it did surprise me a bit because you said that yes about 76% of Malaysians
03:30are very comfortable to let AI compare what they need loans for instance but 57% actually I feel it's
03:39quite high they say that they would give AI agent at least some autonomy in say applying for a credit
03:45card
03:46or a loan that's quite surprising because do Malaysians actually understand at this point what we're saying
03:51yes and no to an AI agent yeah I guess that that could if you look at the personal loan
03:58journey
03:58a typical ones that what we have seen is a genetic AI assisting consumers to apply a loan and the
04:07comfort
04:07level based on the study that we have done is that at the level of when you search and compare
04:14and negotiate the comfort level is above 80% but the moment you get into the application level which
04:22means that you start to say that I will pass my financial data my personal data to the individual
04:28to the agent itself that comfort level will come down to about 60-65% and and the final stage
04:36is that
04:37once they come back it will still roll back to the human so if you look at it even with
04:42this
04:43agentic AI human is still in the loop the final decision goes back to you so depending how do you
04:49define the rules I think that's very important okay so the AI agent needs data to help make recommendations
04:56so let's talk about how much of that am I going to hand over to get good recommendations
05:02for the AI agent to actually decide hey this deal is good for me income spending behavior what kind
05:08of data points are we talking about I guess that in if you look at typically from agentic commerce is
05:14that you tell what the agents to do right for example in the very simple terms how do you leverage
05:21on agentic AI is that I will tell the agents for example the AI hey I want to go for
05:28a three-day
05:29trip going to Bangkok during the Deepawali holiday help me to find hotel that is within the range of
05:37five hundred ringgit and find me allies a reading this timing and help me to book so that is the
05:46thing
05:47how as a consumer you can define the rules device the expectation device the criteria but ultimately the
05:55agents will come back and tells you this is the thing what do you want me to do and in
06:00order for
06:01the agents to act further that come back to the other layer the trust because you got to hand over
06:06your personal data you got to hand over your payment information I think that itself is something is still in
06:13evolution and at this stage as based on the study that we have seen that trust level is still not
06:20as high
06:20as what you have seen as compared to the one that people asking for advice do you think it's a
06:25matter
06:25of time to to have to see that trust level improve the ecosystem is very important the trust infrastructure
06:33is very important before the consumer can gain better trust or even before a financial institutions can
06:41trust that this applications comes in based on what the agents has actually applied for is the trustable
06:49both ways right right is it both way so what we have done so far in u.s. for experiences
06:55that we have
06:56developed the agent trust system ecosystem a framework that we are inviting people from different payment
07:05information identity security as well as infrastructure side come together and we build a trust token
07:12because if you really look at it ultimately the objective is that am i if i'm a bank if there
07:18is a consumer
07:19leveraging agentic ai come to apply a loan am i able to trust that is a genuine consumer
07:26is this a verified individual that goes back to typically how a lot of organizations
07:34compliance sites are looking at know your customer if you heard about this term very frequently
07:39know your customer so nowadays that's why people call kyc but with the agent uh rising now you are
07:48looking at know your agent is this a trustable agent so that's quite interesting as an information
07:53services company how does ai actually change the way you work with your clients many of them are
07:59banks and our credit companies for instance yeah i would say um uh we just had our own innovation
08:07summit uh last week and one of the key common topics people always ask for be it the solutions that
08:15you have the systems that you have they'll ask you do you have ai elements inside that people wanted
08:21to see how this ai can bring the benefits to them but at the same time uh is there a
08:27gut real
08:28is that a governance in place i think that ultimately it becomes very important you can have ai to help
08:35you uh become a lot more efficient automate most of the processes but it has to be operate in a
08:42trusted environment that there is a necessary gut real and safeguard so information doesn't get out
08:49of control that goes back to if you look at what malaysia governments are trying to do
08:54we see we definitely see there's a lot of benefit so there is a national ai uh blueprint for 20
09:0226 to
09:032030 there's a lot more plan there's a lot more work stream going through that and even i think
09:09government is also putting across the ai governance bill so ultimately we all know that ai is here to
09:17see um you need to embrace the ai and how do you make sure that we leverage the ai to
09:23the fullness
09:24right this goes back to trust around agentic ai which is uh like you mentioned this is at the very
09:33uh early
09:34stage at this moment malaysians appear to be very open to let ai do more for them however based on
09:41your own research as well
09:42uh 83 percent are very worried about manipulation fraud as well as impersonations so how do you build
09:50agentic ai or rather agentic commerce uh where people are generally quite worried about using it if
09:57you just look at what we're seeing like scams on social media even so now you add another layer
10:02of ai to it how should people approach this and how are businesses also navigating this
10:08it is not surprising to see that um uh based on the studies that we have done is across 13
10:15market
10:15so malaysia is at the headline level we are we have high trust so as you mentioned uh 76 percent
10:23of
10:23malaysian trust llms to help them to make comparisons but at the same time you also see 83 percent of
10:31them
10:31worry about hey do i get the fake agents is that a fake merchant so is that a hijack
10:38agent help me to help me to do the transaction i guess this is definitely goes back to if you
10:43look
10:43at malaysia itself i think i look at the study um our online scam losses is very huge right it's
10:51just
10:51grown from 1.5 billion in 2024 to almost nearly double to 3 billion in 2025 i think that itself
10:59uh create a lot more fear on individual besides the scam how do i leverage on agent so to us
11:07definitely even from the study that what we have seen a lot of consumer trust the agents or the ai
11:15agent that come from the financial institutions or agents or the institution that they have trusted
11:21so that is also one area as an organization as the institutions as a businesses out there how do you
11:29gain
11:29the trust on the consumer and how do you define and build some things that actually assure the
11:36consumers that this is something that they can use it so what needs to be put in place uh consent
11:40for
11:41protection regulation at this point of the uh growth of agentic ai which is still relatively new
11:48i it i think it takes identity verifications uh consents the trust to all work together because by working
11:57all these things together then you can build a i would say that is a trustable kind of ai network
12:04and if you look at as the digitalization growing we have a pretty high lead digital adoptions in
12:11malaysia market here and that is a very strong foundation for us so continues to see how well do we
12:19build that trust within the ai ecosystem i think that is the way to go it is in the right
12:26directions
12:27with the ecosystem from the government private as well as institutions come together to see what will
12:33be the best way for them to leverage on this i would say that is a transformative kind of technology
12:39okay
12:39now as a customer i'm actually quite happy for ai to help me compare the merits and of five different
12:46credit cards right but but if you do a lot of heavy lifting yeah right i have no time to
12:50read the
12:50terms and conditions so yes ai is great on uh in in that respect but if it's recommending me one
12:57card
12:58over the other how do i know why am i seeing this particular product i i guess that at this
13:07point
13:07most of the lom models are doing their best in terms of based on what is your stated requirement
13:14you may tell them what is my requirement i want a card that can suit my lifestyle my lifestyle is
13:21i
13:21travel a lot so typically they will pick based on what exactly is your requirement and they will find
13:28what are the features that is available through the internet so one of the point is that we know a
13:35lot
13:35of all this purchase behavior loan applications no longer just happen at your website it is actually
13:43happen at an ai conversation just like what you say you start to go to the ai say that i
13:47want to apply
13:48a credit card i want a credit card that can support my travel habit i want a credit cards that
13:53can give
13:54me more reward points this is the criteria they will use it to compare then the questions is that where
14:01do
14:01they find they will search enough of resources and as a user we would always say that how do you
14:07challenge
14:08the ai further you can ask them to quote where did you get the sources of information why do you
14:13make
14:13this recommendation that actually in turn helps you to understand better what goes behind it so the
14:20explainability even through leveraging the ai is equally very important okay so i'm gonna probe a bit
14:25further because as a consumer then i will want to know whether the best recommendation by the ai agent
14:31may actually be influenced by a company that's paying to be on there is that something that you're looking
14:38into in your research right now is it uh our current research is not because that this is just a
14:43rising
14:44trend that we are observing across the market um but as we look at it at this stage pretty much
14:51is based on
14:51discoverability based on clarity based on what content do you put on your website but we definitely there are trends
14:59right
14:59as we see businesses even from social media google other places ultimately would that get to the peak
15:07i think that yet to be seen but uh will be i wouldn't be surprised that would go to that
15:13day ultimately
15:14is for consumer to really um or those by using the ai they must also know what are the recommendations
15:23out
15:24there always come with a mind to challenge the thing if you think is this the best solutions for me
15:31do you have any other things out there the ai agents would definitely helps you to find the best
15:37outcome so that's from the consumer perspective i want to look at from the lens of businesses right so
15:43companies speed in uh credit or banks um let's just widen it out right any company they have spent they
15:49spend years and so much money to try to understand and influence the customer journey but what happens
15:55now is that an ai agent is increasingly sitting in between that layer right with more people using llm
16:01so how does that change the equation how businesses are reaching out to customers and how they're using data
16:06and insights from a company like experience as you rightly pointed out um increasingly the ars become
16:16sitting at the point of financial and commercial decision as what we have seen since at this stage
16:21that businesses are there are a lot of talk as as you can see from the paper uh from the
16:27news is a lot of
16:28organizations are trying to embrace ai a lot of times that is to help them to do the agenda workflow
16:35because that that is the automate automations process that actually they can leverage a lot more
16:40benefits but when come to consumer behavior that is the ones that we are seeing it is still at the
16:47early
16:47stage but it is growing and and from a businesses standpoint financial institutions we are watching
16:55the trend because if this is the trends that is coming in as part and parcel of a very important
17:02transformative kind of technology we have to start to see how do we build in looking at the workflow
17:09working at our processes also even review the products that we have how are we able to be more
17:15visible to the ai discoverability how do we ensure that it is the right recommendations how do we put in
17:24the certain right training because you've got to train the agent as well how do you ensure you feed the
17:29right data point and you put in the right governance to ensure that whatever agents ai products that you
17:37have come out with uh that is actually suited for the market i think that is very important does that
17:42fundamentally change or will change how businesses compete for customers in your opinion in the future
17:48definitely whoever is that conversation that you're having with the uh the clients that you work with
17:54we are definitely seeing that organizations that do not want to lose out they want to be the ones that
18:01understand how they can leverage ai uh in order for for them to help their customers to do a lot
18:07more
18:08better right so in that sense whichever of the organizations is able to embrace ai i i guess that
18:15that would be some things that they will be at the forefront of the competitions in the battleground
18:21itself because i imagine that with ai with agentic ai um the agent doesn't care if i have banked with
18:28a certain company for the last 15 20 years right what it meant what matters is what i need and
18:34these are
18:35the recommendations so what happens do you think to customer loyalty and how banks approach um customers
18:41yeah i guess it's a challenge ultimately um beside just the ai ai is only one of the ways that
18:49how
18:50consumer comes to you uh for a banks for uh organization brand equity is still very important
18:59just as what we have seen from the report even as much as the consumer like to use the ai
19:05trusted ai
19:06the trust level for ai that is ai agent that is built by the trusted organization it is still a
19:14lot
19:14more higher as compared to just any independent ai that they can actually leverage on it
19:19so i would say that ultimately uh customer loyalty how do you engage with the customer it is goes back
19:27to your overall strategy your customer engagement strategy how do you improve your customer journey
19:34how do you make sure that the leveraging ai is to improve the overall seamless process and customer
19:41will find that there are i think trusted agent that is actually assisting them just like you will talk
19:47about chatbot old days maybe a lot of you have tried using chatbot interacting that may not give
19:53you the best answer that you want it with the ai intelligence incorporating in that whether that can
19:59differentiate yourself that can helps you but brand loyalty i would say that definitely go back to
20:04not just purely about ai it is about how you differentiate yourself in terms of your service in terms of
20:11your
20:11pricing in terms of the the customer experiences let me ask you this because experience in different
20:16market many different markets right have you noticed a certain market that appears to have a higher
20:21level of trust to its ai agent making decisions or rather acting on behalf of customers uh we are
20:28currently quite active uh us is our largest market so we are seeing that trend is moving up uh that's
20:35why
20:35that's the reasons that why we felt it is important for for us to set up a agent trust ecosystem
20:42or a
20:42framework that bringing uh various party together because ultimately you need an ecosystem to really
20:50safeguard and making sure this would work well and and that is definitely what we are seeing it now
20:56uh the trust level is from different different market as it is at a different maturity level okay
21:03yeah what do you think when you talk about the shift towards agentic ai what do you think our
21:08businesses underestimating right now or should be preparing for that shift i would say that it is
21:16some things that they would the investment that they needed the skill sets of their own people
21:23uh their internal processes what are the things that they need to change um how do you integrate this
21:30ai processes within your own workflow is one of the biggest challenge and how do you embrace it or how
21:38do you disrupt your own flow to make sure that these are some of the things that you can build
21:43from
21:43scratch and and bring the values back to your customer okay so if you're having this conversation
21:49again say in uh three years um what do you feel could be something that we will be routinely allowing
21:57ai agent to do for us that may be unusual now like just give us an insight into how that
22:01would look
22:01like in the future i would say three years from now maybe by dance that many of us don't even
22:07need
22:07to really look at you don't need to spend a lot of times uh do a lot of booking all
22:16this you will just
22:17entrust your agents to help you to do most of the thing i felt three years is a good time
22:21and this is the
22:23next couple of years for organizations to come together to see how they can leverage on this
22:30technology to actually make every of our life better and as an organization experience how are you also
22:36preparing for that shift if you can share with us uh for us as an organizations um we are we
22:43already
22:43embracing ai be in our own assisting workflow assisting processes but it is also for us look at what this
22:52would be relevant for our clients so we have decisioning solutions we have analytics solutions
23:01that is how we can incorporate the necessary ai tool to actually assist our customers to do their job
23:07better because ultimately is that from a client standpoint do i still need to start from doing
23:14day one from scratch doing things that i need to do or would you have an ai assistant to help
23:21me to do
23:21my job better be it i'm a credit analyst i'm a fraud analyst so do you have an ai tools
23:28to help us
23:28we do build all this ecosystem the different different type of agent to work together to help the banks
23:36to allow them to actually take in some of these agents into their particular workflow i think that
23:42is where we see the directions because ultimately as organization evolve as banks start to embrace
23:49this ai agentic workflow experience as their data and so-called technology partner we are also coming
23:57up such a supporting solutions so we are able to integrate into the banking uh workflow or our clients
24:05workflow all right well we'll probably speak again maybe not three years maybe a year from now and see how
24:10where we are by then sure well thank you so much don for sharing our insights with us and thank
24:13you
24:14for coming on the show thank you thank you