This founder’s looking to another technology leap for AI to start reasoning like humans
In an interview during his visit to Bengaluru where the company set up its first office outside Romania, he spoke about how robotic process automation (RPA) is being used with GenAI to enhance customer experience, and why entrepreneurs will find it hard to survive if they simply build chatbots atop large language models (LLMs). Edited excerpts:
Why did you step down as co-CEO to become CIO? What does this new role entail?
I was lucky to find a great CEO (Rob Enslin) to do the go-to-market and day-to-day operations. Now I have way more time to work directly with the product and engineers, which is my passion. I believe that it’s better to play on your strengths and double down on them.
You scaled your company from a tiny apartment in Romania to a more than $1 billion-revenue public entity, which displays your other strength too.
Agree. But it’s mostly about focus and bandwidth, and especially with the advancements in AI, we are seeing a generational shift in products. It’s much better to be involved in reinventing products.
This would mean using automation not only with AI but with GenAI too…
The shift was in motion before GenAI because we’ve built a lot of our stuff on top of AI. GenAI is adding smarter capabilities to RPA because, in a way, it completes the automation. If you are an insurance company and a customer says: “I have this type of car, please give me a quotation for my car”, we typically enter the data in our systems to get a quotation for the customer.
GenAI can understand the text, so it will understand the request and extract the information. It will then use RPA to access the company’s systems, get the quotation, and use another type of GenAI to reply to the customer, thus completing the entire process without any human interaction.
If the customer accepts the quotation, GenAI will create the customer profile, issue the policy, and email it back to the customer. This is now possible across other sectors like healthcare too. If automation is the body, GenAI is the brain. They need each other to be useful.
What should CXOs consider when implementing GenAI and automation?
We typically ask: What are your workflows? What are your use cases? And then explore how we can automate most of these tasks with automation and GenAI. We have our own GenAI model. For instance, we have a tool called Communication Mining that can read all your emails, understand them, and classify them.
Our models give customers better control of their data because the biggest challenge regarding GenAI is around privacy, security, governance around data. CXOs want to know exactly where their data goes, where it is stored, how it is processed, and who can use it.
We are also working on a tool we call Autopilot to help people using our platform to build faster applications. We also want to use LLMs to make our automations more reliable. Automation uses different systems, and we plan to use LLMs to better understand the context of small changes and accommodate them on-the-fly.
In the longer run, we are working on a foundational digital assistant model where business users can talk to a model or interact using text, following which the model can understand and act.
How much work is being automated in companies? And how does one re-skill or upskill people being replaced by automation?
It’s still early days, but in some use cases we can reduce manual interaction by almost 80%.
The reality is that there is more work than there are people capable of doing that work. Automation didn’t increase unemployment. On the contrary, we are seeing a big shortage of labour because most nations have an aging population with more people retiring and fewer people entering the workforce.
Further, the new generation is not as willing to do the same jobs as the people who retired. So automation is clearly the only way to cope with the problem of labour shortage.
You typically automate routine and repetitive tasks, and not the job itself. This helps you focus more on meaningful jobs rather than mundane tasks. For instance, during the covid pandemic, we worked with hospitals to automate some registration for covid tests that involved a lot of administrative work. This helped in reducing queues and was very helpful since there’s a shortage of nurses.
Agree. But given the incredible pace at which AI and GenAI are automating tasks, what should employees consider when upskilling themselves?
The main difference between people and everything else, including technology, is that we can (still) distinguish between truth and non-truth. And this is the biggest problem of LLMs today—they cannot distinguish, they hallucinate but have no idea that they are doing so.
In the longer run, I believe we will have to completely change our education system. Because our education system was devised about 200 years ago and mostly to cater to people who worked in factories and do repetitive work. We must go back to the basics and make sure that people are in control of the technology.
This is easier said than done. How, according to you, should we go about this task?
I use ChatGPT daily, but mostly as a search engine since it helps me find some answers faster. But it reduces the code writing time for developers, thus increasing their productivity. This is where Microsoft Copilot was so successful, and this is where I feel lies (GenAI’s) biggest impact—reading documents, understanding, and extracting information. But to productise (ChatGPT) in the enterprise context is proving to be difficult (since GenAI tools still can’t reason).
There are two camps of thought here. One says you just put more nodes in the network, and at some point, (GenAI) will become extremely smart. The other camp says there must be another giant leap in technology. I am more in this camp that believes we must create another big innovation to make these models reason like humans.
Having built a successful company, what’s your advice to budding entrepreneurs who are trying to build an AI or GenAI startup?
I’m seeing hundreds of companies, especially in AI, that don’t bring any value. They just add a thin layer on the top of GPT. This is not going to work because the provider (in this case, OpenAI) is always capable of introducing more advancements—like OpenAI introducing its GPT store and putting a lot of startups out of business.
For me, if you are in AI, you must be a great researcher who can create a company that can do serious work in foundational models. If you are not in this camp, you will have to, in my opinion, find a very specific area that can help a customer in a particular domain.
In an interview during his visit to Bengaluru where the company set up its first office outside Romania, he spoke about how robotic process automation (RPA) is being used with GenAI to enhance customer experience, and why entrepreneurs will find it hard to survive if they simply build chatbots atop large language models (LLMs). Edited excerpts:
Why did you step down as co-CEO to become CIO? What does this new role entail?
I was lucky to find a great CEO (Rob Enslin) to do the go-to-market and day-to-day operations. Now I have way more time to work directly with the product and engineers, which is my passion. I believe that it’s better to play on your strengths and double down on them.
You scaled your company from a tiny apartment in Romania to a more than $1 billion-revenue public entity, which displays your other strength too.
Agree. But it’s mostly about focus and bandwidth, and especially with the advancements in AI, we are seeing a generational shift in products. It’s much better to be involved in reinventing products.
This would mean using automation not only with AI but with GenAI too…
The shift was in motion before GenAI because we’ve built a lot of our stuff on top of AI. GenAI is adding smarter capabilities to RPA because, in a way, it completes the automation. If you are an insurance company and a customer says: “I have this type of car, please give me a quotation for my car”, we typically enter the data in our systems to get a quotation for the customer.
GenAI can understand the text, so it will understand the request and extract the information. It will then use RPA to access the company’s systems, get the quotation, and use another type of GenAI to reply to the customer, thus completing the entire process without any human interaction.
If the customer accepts the quotation, GenAI will create the customer profile, issue the policy, and email it back to the customer. This is now possible across other sectors like healthcare too. If automation is the body, GenAI is the brain. They need each other to be useful.
What should CXOs consider when implementing GenAI and automation?
We typically ask: What are your workflows? What are your use cases? And then explore how we can automate most of these tasks with automation and GenAI. We have our own GenAI model. For instance, we have a tool called Communication Mining that can read all your emails, understand them, and classify them.
Our models give customers better control of their data because the biggest challenge regarding GenAI is around privacy, security, governance around data. CXOs want to know exactly where their data goes, where it is stored, how it is processed, and who can use it.
We are also working on a tool we call Autopilot to help people using our platform to build faster applications. We also want to use LLMs to make our automations more reliable. Automation uses different systems, and we plan to use LLMs to better understand the context of small changes and accommodate them on-the-fly.
In the longer run, we are working on a foundational digital assistant model where business users can talk to a model or interact using text, following which the model can understand and act.
How much work is being automated in companies? And how does one re-skill or upskill people being replaced by automation?
It’s still early days, but in some use cases we can reduce manual interaction by almost 80%.
The reality is that there is more work than there are people capable of doing that work. Automation didn’t increase unemployment. On the contrary, we are seeing a big shortage of labour because most nations have an aging population with more people retiring and fewer people entering the workforce.
Further, the new generation is not as willing to do the same jobs as the people who retired. So automation is clearly the only way to cope with the problem of labour shortage.
You typically automate routine and repetitive tasks, and not the job itself. This helps you focus more on meaningful jobs rather than mundane tasks. For instance, during the covid pandemic, we worked with hospitals to automate some registration for covid tests that involved a lot of administrative work. This helped in reducing queues and was very helpful since there’s a shortage of nurses.
Agree. But given the incredible pace at which AI and GenAI are automating tasks, what should employees consider when upskilling themselves?
The main difference between people and everything else, including technology, is that we can (still) distinguish between truth and non-truth. And this is the biggest problem of LLMs today—they cannot distinguish, they hallucinate but have no idea that they are doing so.
In the longer run, I believe we will have to completely change our education system. Because our education system was devised about 200 years ago and mostly to cater to people who worked in factories and do repetitive work. We must go back to the basics and make sure that people are in control of the technology.
This is easier said than done. How, according to you, should we go about this task?
I use ChatGPT daily, but mostly as a search engine since it helps me find some answers faster. But it reduces the code writing time for developers, thus increasing their productivity. This is where Microsoft Copilot was so successful, and this is where I feel lies (GenAI’s) biggest impact—reading documents, understanding, and extracting information. But to productise (ChatGPT) in the enterprise context is proving to be difficult (since GenAI tools still can’t reason).
There are two camps of thought here. One says you just put more nodes in the network, and at some point, (GenAI) will become extremely smart. The other camp says there must be another giant leap in technology. I am more in this camp that believes we must create another big innovation to make these models reason like humans.
Having built a successful company, what’s your advice to budding entrepreneurs who are trying to build an AI or GenAI startup?
I’m seeing hundreds of companies, especially in AI, that don’t bring any value. They just add a thin layer on the top of GPT. This is not going to work because the provider (in this case, OpenAI) is always capable of introducing more advancements—like OpenAI introducing its GPT store and putting a lot of startups out of business.
For me, if you are in AI, you must be a great researcher who can create a company that can do serious work in foundational models. If you are not in this camp, you will have to, in my opinion, find a very specific area that can help a customer in a particular domain.
Denial of responsibility! Techno Blender is an automatic aggregator of the all world’s media. In each content, the hyperlink to the primary source is specified. All trademarks belong to their rightful owners, all materials to their authors. If you are the owner of the content and do not want us to publish your materials, please contact us by email – [email protected]. The content will be deleted within 24 hours.