← The Digital Executive
The Digital Executive
Automating Knowledge Work: Alberto Rizzoli on Building Trustworthy AI and the Future of Work | Ep 1140
00:15:55
Loading summary
Transcript13 lines
- [00:00]
A
Foreign welcome to Coruscant Technologies, home of the Digital Executive podcast. Do you work in emerging tech? Working on something innovative? Maybe an entrepreneur? Apply to be a guest at www.corazon.com brand welcome to the Digital Executive. Today's guest is Alberto Rizzoli. Alberto Rizzoli is a serial entrepreneur and AI systems expert, best known as the co founder and CEO of V7, a UK based company building AI to automate knowledge work. At V7, Alberto helps enterprises like GE, Healthcare, Merck and Sony streamline their most document heavy and detail driven workflows, turning unstructured data into structured insights that fuel efficiency and scale. His work focuses on practical AI that delivers measurable outcomes in high stakes environments from healthcare to finance to insurance, where accuracy and compliance are non negotiable. Before B7, Alberto founded IPoly, a pioneering computer vision app that identified objects in real time to assist people who are blind or visually impaired. The technology scanned over 2 billion objects worldwide and was translated into 26 languages earning Alberto recognition from the President of Italy and multiple CES Best of Innovation awards. Well, good afternoon Alberto. Welcome to the show.
- [01:26]
B
Hey Brian, great to be here. Thank you for having me.
- [01:29]
A
Absolutely my friend. I appreciate you making the time in London, England. I'm in Kansas City, so got a six hour difference today, but I appreciate you navigating the time zones. And Alberto, I'm going to jump into your first question. You built AI Poly to assist visually impaired users and now lead V7 to tackle enterprise AI workflows. What were the hardest lessons you learned in transitioning from a mission driven consumer product to the B2B infrastructure platform and how do they inform your decisions today?
- [01:59]
B
Yeah, so Ipoly, my, my previous AI company that I started in 2015 was very mission driven. It was helping some of the least privileged people in society, which are people with total or almost total visual impairment. So the totally blind effectively. And it was using AI to identify objects in real time from your phone without requiring an Internet connection. So entirely using the processing power of smartphones available back in 2015, which were not very powerful at all. Neither was AI very powerful, but it was very mission driven effort. And as mission driven as B7 is today, because its impact on the healthcare industry, on insurance actually leads to massive improvements in life. It feels less obvious because it Ipoly would be an app that would get the users who tried it the first time to be moved and sometimes moved to tears, which was incredible. It had this really strong viral approach and most importantly was a consumer app so anyone could download it on their phone at the dinner table, test it out with their phone around and would speak back at them saying, hey, I see a plate of steak and I see a wine glass and it would be really fun to do. Well, V7 on the other side has significantly higher impact. It was founded in 2019 and it automates anything that we consider to be back office work. So all the paperwork processes that large companies need to do are now handled by the seven AI agents. And we've been able to manage hundreds of billions worth of transactions that would normally have to be done manually and paperwork that really have to be done manually that ends up becoming a huge cost to society. 40% of our administrative of our costs in the health care system in the uk, the NHS is administrative. It's people that are not helping the patients on the ground. They're not nurses, they're not doctors, they're not surgeons, they are in the back office handling effectively the ineffective bureaucracy of that system. And you need to develop a very strong sense of storytelling in order to inspire people to solve some of the most ugly and boring problems in society. And you also need to make it fun. So something that we focus a lot on in developing D7 is making the user experience of the product as fun and as intuitive as possible. So you feel like you're orchestrating these automations as a designer of this workflow, as opposed to doing the actual boring work that these automations, these AI agents eventually tackle. So the main lesson is if you're building something B2B, you have to make it fun and you need a really good storytelling angle to it so the people are inspired by the mission behind it. As much as B2B missions tend to be much more stale than consumer ones, especially ones in assistive technology like High Poly was.
- [04:52]
A
Thank you so much. I appreciate that and love the backstory. Obviously you're very much mission driven. Your original business was to help people with disabilities, serving consumers, and now you're doing the same thing with the E7. We're working to help humanity in the healthcare space, leveraging the latest AI technology. But I also highlighted your good storytelling makes a difference and obviously making things fun is always goes a long way, so I appreciate that. And Alberto, V7 emphasizes combining human expertise with AI to deliver trustworthy AI. How do you operationalize that in practice, especially in high regulated domains like healthcare, finance, so that clients feel confident in automation decisions?
- [05:38]
B
Yeah, effectively. AI knows a lot about everything, but it often doesn't know how to do your job or the parts of your job that you want to automate well enough for you to actually trust it. V7 allows you to show an AI the step by step process of anything that you do. And it lets you explain to it how to handle any edge case for it, so that ultimately you're creating your own AI agents that are almost like your own employees, that are able to tackle these in a way that is reliable. And for anything that is an edge case, they can come to you and ask you for questions and clarifications. The step by step development of a workflow is essential to making AI actually feel like something you can rely on. In healthcare, in health insurance, in. There's a lot. There's no margin for error really. And there's also no tolerance for automation errors. Much lower. We're much more forgiving if a human makes a clerical error. And there are many of them in especially regulated industries. But we're very inflexible towards automation because it makes us feel like we've gone the cheap route and we have therefore created a mistake that is systemic and it's affecting people. And so it's very important to develop software that can reach as high of an accuracy as possible and is reliable in the way that it discloses its errors. The way we do it in practice at V7 is that every decision that the AI makes and every piece of analysis is forced to be grounded in documented evidence. That means that if an AI is making a decision based on someone's policy or based on someone's patient pathway, or based on someone's. We also process mortgages, for example, which is a really high impact activity because it's, it's literally someone's future on the line and their home. It always needs to provide not just a source evidence as a link, but it needs to go into a document and find the exact statement and then highlight it with a box so they can never hallucinate. And it's always giving you a traceable answer. And it's also developed to not be overly optimistic about what it wants to tell you, to be very factual with the presentation that it gives you back, and not to try and make the user happy like many AIs do, and to ask for help for anything that is subjective and requires a bit of a human opinion to it. And there, there's a lot of technical detail that goes into this. But if you want to learn more, v7labs.com is a great place to go check it out.
- [08:13]
A
That's awesome. Thank you. And I agree with you. AI has come a long way I think it still has a long ways to go. And your platform aims to stand out in the industry because your platform can follow specific job tasks and mimic day to day workflows. And your platform does go beyond automation. It actually looks to logically solve issues. And what I really liked, Alberto, is it asks for help for anything that is really subjective or needs some clarification. I think that's really important. And Alberto, when your platform is used in healthcare, pharma or insurance settings, errors or biases can have real consequences. What governance, validation or oversight frameworks do you insist on inside V7? And how do you partner with clients to enforce them?
- [08:59]
B
Yeah, so recently certifications have been a lot more streamlined for software companies. There's things like ISO 27001 and SOC2 and all these large checklist exercises that help you effectively know that the company that you're dealing with, the vendor you're dealing with, AI has data security and integrity. But I think getting to know the people behind the software is quite important. Knowing that they are not vibe coding their way through a piece of software, but actually building something incredibly robust. And often that's quite hard. When selecting an AI vendor, we often get asked whether any of the data that we process ends up being used to retrain models. We get asked that almost all the time. And it really shows that there's a concern of data leakage. There's a concern that AI will ultimately use your information to improve itself. And we absolutely don't do that. However, there is still this sense of hey, we are delegating everything to this AI. Couldn't it not just understand everything about our business and then effectively replace what we do? We've decided not to take a stance of going after our customers, of course, but you know, they know their area best and we want to provide them with the infrastructure to develop agents that belong to them, prompts that belong to them, output data that belongs to them. And this was really important for us from the start because we can't really have a. We're all about trustworthy AI. We can't claim to be creating trustworthy AI if we ourselves as the vendor cannot be trusted with our customers data. So it's a very sacred element of what we're given. And a lot of AI selling is trust building, is making sure that the customer on the other side sees you as a partner for the long run and understands that we are here to help them build something incredible with AI and not to effectively be this vacuum of data which has happened in the past with Some software companies and I think the. We're at the stretching limit of people's patients with the information not being treated correctly.
- [11:10]
A
Thank you. And I really appreciate that a lot of things nowadays are more streamlined, more efficient as far as certifications, whether you're hooking into an API, that sort of thing. But I liked how you explained it is important to you that you build your trust with the customers, show them that you're there to help and to be a long term strategic partner. I appreciate that. And Alberto, last question of the day. Looking ahead maybe 5, 10 years, how do you hope V7 and your work will change how we think about intelligence, knowledge, work and human augmentation? What legacy do you want to leave not just in tech, but how we live with AI?
- [11:49]
B
Yeah, things are going to be wild in five years. I think the rate of improvement of AI is slowing down a little bit. But there are still enough gains for us to capture to see the world very much transformed. I think we will see more changes in the next five years than we have seen in the last 15 and even the last 15. You know, we didn't have the smartphone at the time, so there's been some really big changes. And our legacy is in transforming what we consider to be work. We've created a society where after the industrial revolution we've moved into big cities so that we could do this new. We call it the tertiary sector in Italy. I'm not sure if it's saying is in English, but it's services or the service industry that for example, the United States considers to be its biggest contributor to GDP is the new agrarian industry pre industrial revolution. It will change significantly, but it will not necessarily make things more or less fun. It will remove and I hope that Vs. Emma will remove all the bureaucracy that we've created in order to systemize the services industry, whether it's legal services, insurance services, financial services. And it will also reduce the margin of knowledge arbitrage to near zero, which means that our insurance policies should become cheaper because we no longer have all this administrative cost. Our public funds should be used more wisely because we have, we have knowledge workers on tap. We have effectively the power of administering a budget as if we had unlimited time to spend on it rather than having to rush through the allocation of funds which often lead to painful mistakes and the mismanagement of taxes. And the societal changes that we will see will first of all lead to a change in labor where we will have fewer individual contributors in the domain of knowledge work. So fewer paralegals for example, or juniors, which is will kind of have a negative impact on young graduates. But these young graduates will want to adapt to becoming workflow designers and AI designers effectively. And it's becoming accessible to even people that don't know how to code to be able to look at a process and turn that process into an AI enabled one where humans do less than 20% of the work. I think that's a huge opportunity for young people because this type of skill is in demand everywhere around the world and in almost every company. And then for the remaining part, the work that we don't automate and don't want to automate is everything that is interpersonal and creative. And I think it's the sort of core of the human existence is to use our creativity and our social skills to advance a mission. And I hope that we will remove everything that is considered a, a follow up that needs to be done or an administrative scheduling task or the management of your taxes is not the reason why we're on this earth. And I hope that that's something that will offload to AI very soon.
- [14:54]
A
Thank you so much, I appreciate that. Just to highlight a couple of things, Alberto. AI may be unrecognizable in the next five years. We know that it is leapfrogging. But we hope that society, humanity is better because of it. And we. You also mentioned your hope is your platform B7IS will help be there to help remove some of that bureaucracy. And I did highlight something that I thought was important is you see an opportunity in the younger generation to actually leverage AI to maybe help tackle a new process or improve your process. I thought that was really insightful, so I appreciate that. And Alberto, it was such a pleasure having you on today and I look forward to speaking with you real soon.
- [15:35]
B
Thank you Brian. Great to be here. And for anyone that is young and wants to automate a process, come talk to us@thev7labs.com before Victoria. Have a good day.
- [15:46]
A
Bye for now.