Timestamps

(00:00) Introduction;

(00:29) How a tech exec from China joined a Singapore AI startup;

(02:38) The evolution of LLMs in China;

(03:33) AI Development in China and the rest of the world;

(04:37) The value of building a career in China tech;

(06:34) Scaling a global AI company;

(08:37) Scaling in Southeast Asia vs rest of the world;

(09:39) Driving the AI conversation and building strategic relationships;

(10:34) Future of Conversational AI;

(13:02) MCPs impacting the cost structure of building AI;

(14:26) New paradigm of pricing AI products and solutions;

(15:00) WIZ.AI’s AI Roadmap;

(15:54) Considerations of Enterprise Buyers;

(17:14) AI Enterprise Sales;

(22:30) Advice on AI Transformation;

(24:17) Make or Break Moment;

Follow us ⁠on LinkedIn⁠ for more updates

Subscribe to ⁠our monthly newsletter⁠ for all the news and resources

Directed by Paulo Joquiño

Produced by Paulo Joquiño

The content of this podcast is for informational purposes only, should not be taken as legal, tax, or business advice or be used to evaluate any investment or security, and is not directed at any investors or potential investors in any ⁠⁠⁠⁠⁠⁠Insignia Ventures⁠⁠⁠⁠⁠⁠ fund. Any and all opinions shared in this episode are solely personal thoughts and reflections of the guest and the host.

About the Guest

Robin Li is the Senior AI Strategy and Partnerships Director at WIZ.AI, bringing over 20 years of enterprise technology experience and more than 10 years specializing in AI companies. Based in China before joining WIZ.AI in Singapore, Robin has witnessed firsthand the evolution of AI development across different markets, from the early days of conversational AI to the current LLM revolution. At WIZ.AI, he focuses on bridging the gap between AI builders and enterprise buyers, helping organizations navigate the complex journey of AI transformation while building strategic partnerships across global markets.

WIZ.AI is a conversational AI talkbot platform for 300+ global enterprises present in 17+. countries. Their conversational artificial intelligence talkbot mimics conversations with real people by localizing for 17+ different languages and dialects from ASEAN to Latin America. The company’s tech solution replaces traditional human call-centers for different application scenarios (e.g. tele-sales and customer care) across a variety of sectors (financial services, telecommunications, healthcare, etc.). The company’s enterprise grade proprietary technology is built on the company’s R&D into natural language processing, automatic speech recognition and text-to-speech technology. WIZ.AI is a first-mover in untapped emerging markets, with 11 patents in conversational AI technology. The founding team cumulatively has over 40 years of experience in FinTech, big data, artificial intelligence and cybersecurity.

Transcript

Introduction and Guest Welcome

Paulo: Thank you very much, Robin for coming on call with us. I’m really interested to learn more about your experience coming from many different tech companies in China, and then eventually coming here to Singapore to work with WIZ.AI on what they’re building, which is not just the Southeast Asia solution for conversational AI, but really a global one. They have clients all across the globe, including even Latin America. Quite an interesting company and always great to hear from their leaders. I think we’ve had Jennifer a couple years ago on the show and now it’s great to meet another leader at WIZ.AI. Maybe you can do a quick intro of yourself for our listeners and for our audience, and then we’ll go from there.

Robin’s Background and Journey to Singapore

Robin: Thank you, Paulo. I’m Robin from China and I have worked for AI companies for about more than 10 years. I have worked in the enterprise technology industry for more than 20 years. Actually I knew WIZ.AI for a couple years before, and what I know is that I got the journey about WIZ.AI, that they did very well on the voice board and also have a very strong customer base in Asia, this region.

I think that was around the Dragon Boat Festival this year. James came to Shanghai and discussed with me about the new vision about WIZ.AI, about the AI partnership in the area. That effectively caught my interest. We discussed a lot about how the large language model is changing the game in enterprise technology and also we exchanged a lot of insights in what we know, comparing between Asia and mainland China.

That led to the question where he asked me if I’m considering an opportunity in Singapore. That’s why I’m here because to me that was a very natural decision and very quick decision making. You consider the macroeconomic environment, you consider the AI trend in Asia, the company’s interest in AI solutions, and also my positive impression of Singapore, so I came here.

In the past two months, I met a lot of customers and partners, and it’s a fantastic environment that is encouraging for customers about AI. Everyone is very willing and also has a very good culture to embrace and understand AI now. That’s really good here.

Paulo: So you mentioned a couple of interesting things. One is that you talked about the development of LLMs. You’ve worked in a couple of interesting companies back in China that are already doing AI and all of that. What are your views on how the LLM space has evolved in the last few years?

Discussion on Large Language Models (LLMs)

Robin: I think from two years ago we saw that LLMs started from the “aha moment” from ChatGPT. A lot of companies, not only startup companies but big companies, are changing their AI technology involvement. In China especially, in the two years you have all seen that they have accelerated their pace about how to adapt this kind of technology into applications.

I think that is very interesting and I saw that a lot of companies, a lot of industry companies are trying to use LLMs to change their workflow, even changing their business models.

Paulo: Another interesting thing you mentioned about talking to James is the difference between AI development in China versus the rest of Asia. Maybe you can share a little bit of what you talked about with James and especially compare that also to other parts of the world. What are the differences in terms of how AI is developing?

AI Development in China vs. Southeast Asia

Robin: In China right now there’s a top-down trend that all of the leaders are trying to push the new technology into their business because they’re facing more challenges about business growth and also about cost saving. But in Asia right now, companies normally, the industry people and customers, they’re waiting for the new trends from the AI technology startup companies.

They need the information from companies like Google AI, the platform companies to get the information together and they’re waiting. They have a lot of business cases they want to use AI for, but right now there’s no existing AI solution that fulfills their needs here. Also they see a lot of products designed for China, for America, but not here.

Paulo: I guess that’s what really attracted you to WIZ.AI because it’s a product designed specifically for, initially designed for Singapore companies, for Southeast Asia companies. Could you describe a little bit how your past experiences working in tech in China have shaped how you’re approaching your role now with WIZ.AI?

Conversational AI and Its Evolution

Robin: Right now I think we can talk about conversational AI. That’s what WIZ.AI is focused on. To me, conversational AI is not only a piece of technology but also a benchmark or even itself is a very core goal of artificial intelligence. Because back to the story, you know the famous Turing test, that is a conversational scenario. We are supposing that the computer or AI can persuade the people facing it, conversing with it, that it cannot be distinguished as a human.

What I see is that conversational AI right now has developed very fast based on LLMs, the large models, but we need to notice that conversational AI is not new, it’s long before the large language model. Just based on ASR, TTS and NLU, these kinds of algorithms, it has already gained very steady and really useful business teams. Just like the WIZ.AI solution, we already have gained a lot of scenario use cases in our industry.

If they can be used in not so open use cases, it’s not so open conversation and it is not quite so human-like conversation intelligence, but this already gained business value and it has scaled into the industry. That’s what I can see.

Paulo: Quite interesting because I think LLMs are just the latest development in conversational AI. You guys have been, WIZ.AI has been running since 2019, I guess long before all these LLMs went into the mainstream. What can you share about how WIZ.AI is scaling out as a global company and how is it developing a competitive moat against, especially today, there are a lot more AI companies around and building what you call wrapper companies, building around existing LLMs and all this?

Scaling AI Solutions Globally

Robin: You are right. There’s quite a challenge that we see, a lot of competitors coming here and I think we see that as a positive signal because that means you know there’s business value here. I think the total difference is that we are not right now starting from zero. We started already with fundamentals about the solution and also the business case. Right now, when the large model technology is here, it helps us to attend more use cases and to us that is market expansion.

But I think to most of the competitors, the new company startups, they are right now just working on how to adapt technology into new use cases. That’s a total difference. We have a massive head start. That’s what we see in this market, and right now, our GTM strategy is not focusing on the top. You know we have a lot of customers with big names right here in the region. I think they are quite innovative in AI right now. What we are doing is trying to dig with them about the use cases. How do you get the most benefit and most ROI and also measurable value?

When we have matured in this case, the solution, the methodology, and also the change management in this kind of tool, we go abroad, we go to another market and go globally. That’s our strategy right now.

Paulo: Speaking of going globally, what are the learnings so far from how different it is trying to scale in Southeast Asia versus say maybe Latin America or other parts of the world?

Challenges and Strategies in AI Implementation

Robin: I think there’s quite a lot of difference between different markets right now. I think it’s a challenge but also a positive point that we can see. Here in Asia, different countries have different requirements and also have different governance about data, about hosting, how to host your server, these kinds of things.

That is a barrier or it’s the most complicated point for big companies. But actually for our startup, like WIZ.AI, we are agile and also we are adapting and we grew up regionally. Here we can get the difference and we can know what is the most efficient application scenario and what can we get the most benefit from the good data infrastructure right now that we’re building in these countries.

Paulo: Speaking of getting the benefits from the different local infrastructure, your role is really trying to build these stakeholder relationships. How do you learn from that point of view as somebody who’s trying to talk to the government, talk to other business leaders and things like that? How are you able to manage those relationships and make them really benefit with WIZ.AI?

Robin: I think right now we have a strong sales team and also we have partnerships. A lot of partners are helping us talk with these companies and also the government. That includes, you know, in the past two years we have done some projects for the government, very complicated cases about how to do online services for customers and for the government. Also we did chatbots in the nation-hosted area and we get a lot of efficiency effects here. There is a lot of pride there.

Future of Conversational AI

Paulo: I’m interested to know what do you see as the future of conversational AI moving forward. LLMs are just the latest in a host of different technologies that have pushed conversational AI forward, like you mentioned TTS and all these other things. What’s the next step for conversational AI?

Robin: Right now we are using the large language model to benefit our fundamental product. For example, it will be covered in the long tail about use cases that require more intelligence. For example, the inbound call where the customer comes in, you don’t know what the purpose is originally. It’s a very open scenario. The large model will help us more. Our product right now covers a lot of new use cases, that’s the first step.

The second step is that we found the large language model will help us to enable all the small language models from our TTS and ASR to have the training, have the material and the conversation material, these kinds of things, and to strengthen our process, how to do the onboarding process for us.

Second, we saw that with the growth of conversational AI and also how it’s boosted by the large model, a lot of capabilities will be boosted up. For example, how you can handle the conversation with the customer when he sends you an image. This is a totally new use case we didn’t mention, we didn’t meet before or we were not required to handle that before. But here we are trying to emerge from all this kind of evaluation, all of the technology growth and also embed it into our system. That is all the roadmap that we can see in voice boats or we can see in chatbots in general right now.

Paulo: So it’s opening up different use cases, making it more flexible, easier to build new scenarios. Or even, I like what you mentioned about inbound calls, the ability of the chatbot to just assess already what the scenario is without having to be built for a specific scenario.

Robin: Not so many boundaries about conversation content.

Paulo: Instead of having a collection chatbot or a customer service chatbot, it’s just one chatbot that can do multiple kinds of use cases.

Robin: And then practically that would have some agents that you can call, and I can transfer to another agent.

Paulo: One agent and then another layer of different agents. Quite exciting. How do you see, I think there’s a lot of talk right now about MCP models, context protocols and how it allows developers to be able to offload costs to the large language models and not have to spend so much to build internally, what are your views on that?

Robin: It’s a very interesting question because the pricing model has totally changed for AI, comparing AI applications with traditional software. What I see is that they are charging by the token cost or even by how many agents will replace the human. I think I have some opinion that this is just temporary to me, because what I see is that if you are calculating about the cost, comparing about the human, that means we are assuming that AI is right now changing the existing workflow.

I think it’s not the future change because that is a fundamental change. The workflow will be changed, the organization will be changed, and even the responsibility of the human will be changed. I think it’s not a simple comparison about how you can replace a head cost. I think in the future, maybe we are considering the business value in a new measured way, but not just comparing costs to humans or the token cost model. I don’t think so.

Paulo: So how do you think that will affect pricing for a lot of these AI companies and the business model even?

Robin: I think the token cost way is very beneficial for startup companies because it allows us to use the AI infrastructure at very low cost in the first step. We can invest a little to go into this area and to make some applications, make the prototype.

Paulo: So that’s good for right now. It allows you to run a lot more experiments quickly and cheaply as well. How do you see WIZ.AI in terms of, I remember two years ago, Jennifer was talking about her vision of AI products being really flexible and very user personalized solutions. Where is WIZ.AI in terms of heading towards that kind of vision?

Robin: We saw that already changed a lot, that benefited from the large language model. Right now, just like you said, we can set up different personalities and also different languages and also about gender, about a much better customer experience right now. What we can see is that maybe we can make that distinction for each one, just like you see ChatGPT. Very personal AI, personal assistance, like this way, I think that will be happening.

Using some advanced features, it’s really going to make us more confident to make the onboarding process to the end users. That’s something different right now.

Paulo: What are the other remaining challenges when it comes to selling to enterprises? What is something that stops them from choosing one AI solution over another? Are there still barriers to entry in that case?

The Builder-Buyer Gap in Enterprise AI

Robin: You mentioned the builders and buyers. That reminds me that I just read a report from MIT that’s called “AI and General AI in Business in 2025.” It mentioned that generative AI is divided. It mentioned that they have the key finding that 95% of organizations are getting zero return from their AI efforts. They’re trying to find the gap. They studied about the builders, which are more about the startup companies providing AI products, and they studied about the big companies, the buyers.

They found that the builders are successful because their AI products are always focusing on narrow use cases and try to get success in just those use cases. But the buyers, the big companies, the enterprises, are always asking for more complex solutions. Right now they’re asking for products that can be filled into their specific workflow, integrated with their existing tools, and then considering their employees’ habits and behaviors, and also they want to have data boundaries.

That’s a big gap. That big gap is just our AI partnership opportunity. That’s what we want to build up to cooperate with our customers, that we want to bridge this gap. We know that they are not only needing a static AI product, but also a long-term partnership that can help them to make the change, help them to adapt to the AI era and help them to go into change over time.

That’s what we want to provide. That’s your role. We want to build up that because if you look back to see what WIZ.AI can provide, we are already involved very deeply into the customers’ workflow and building. We also built up on AI technology from the very early conversational AI and now with large models, we know each other and also we found that we have, and we are building right now and we already have something. We have the tools to evaluate the business value. We have the methodology to help make the change, and we have some products already there. This is what we can do.

Paulo: Quite interesting. I guess a lot of you have the WIZ.AI, the AI product itself. But then you are also working with them, especially the larger enterprises. Do you see these AI partners, quite unquote, also having agents for that as well instead of like a customer success team, like they would also work internally with the platform?

Robin: Of course. Because that’s what we see with our customers. I think there are two different kinds of drivers about how to have AI. One is ROI driven. They’re very focused on how to do the use case, how do you help me to do the project to deal with this use case and measure that with the running cost and one-time deployment cost, these kinds of things.

Second part, they’re not only focused on one business case, they want to build an entire AI architecture for all departments. They want to have the capability to build agents for different parts, like customer service and internal studies, some reporting, analysis and maybe governance agents. We help them to make the AI foundation. I call that AI infrastructure build up and also evaluation framework.

Paulo: Before, when it comes to SaaS, I guess sometimes a lot of these onboarding would take several months, sometimes even a year. With AI, how is the turnaround usually? How is it impacted?

Robin: I think it will be accelerated right now, but it is almost the same methodology. We do some POC to prove what value we can get, and then we decide about the solution, then what we need to change, what systems we need to integrate with, and then we go to the end solution. The most major steps are the same.

But right now with LLMs, the POC will be much quicker. Also we can use our experience to make the agent to help us make the methodology done, get the solution very quickly for the end-to-end solution. Also, you know that AI coding is very strong now for interface integration.

Paulo: Do you know how much faster it is now to turn around?

Robin: I cannot measure the average, depending on complexity, but what I saw is that all of our programmers are already using AI coding right now, and I think that maybe 1.5 to 2 times the capability.

Advice for Leaders on AI Transformation

Paulo: As we get to the end of our chat, I wanted to ask for your advice for leaders who are listening in, who are thinking about AI transformation as well. What advice do you have for them as they’re maybe experimenting with a lot of different AI tools and worrying about spending too much on all these different AI solutions? How should leaders think about experimenting, spending on these solutions and getting the ROI that they want?

Robin: I think right now nobody’s doubting about what AI is. I think right now the question is how you can adapt to AI as early as possible. The first thing, I think for all the leaders, for industry enterprises, they should be thinking about AI architecture rather than the technical architecture right now existing. They should be thinking about how to balance between different foundational models. From the enterprise perspective, we never want to depend on one vendor.

Also I think we can gain AI evaluation step by step from the easy cases, from the most ROI we can get from the easy cases. From the successful AI use case deployment, we can go to another one. Just step by step, that should be okay.

I think the best thing for all the leaders to think about is how humans are working with AI, because that is a question even for us. We are thinking about when AI is coming, when AI is stronger and more capabilities are increasing, the use cases are increasing, what is the exact workflow right now? What is the person the company needs immediately? Everything is changing. I think leadership should be thinking about that.

Robin’s Career Challenges and Closing Remarks

Paulo: Finally maybe you can share with our audience what has been the most challenging point in your career so far that other leaders can learn from?

Robin: I think it’s a big challenge that I went overseas from fast-growing China to see the world, because I think the world is changing and the changing is because of the US and China conflict issues and also the AI technology innovation. These two strong factors make everything changing. I think you should keep in mind to learn, to see what is happening and to adapt to what you are doing right now.