AI for Good Innovation Factory Grand Finale 2025

Summary

This transcript captures the finale of an AI for Good Innovation Factory competition, where four startups presented AI solutions addressing global humanitarian challenges. The event was moderated by Peter Vanham from Fortune, with three judges evaluating the presentations: Seizo Onoe from ITU’s standardization bureau, Ebtesam Almazrouei from the UN AI for Good Impact Initiative, and Werner Vogels, Amazon’s CTO. The judges emphasized looking for widely deployable solutions serving marginalized communities and addressing global challenges with scalable, affordable innovations.

The four finalists each had two minutes to present followed by three minutes of judge questions. Elevate AI Africa presented Mama Mate, an offline AI-powered device costing $25 that provides WHO-approved maternal and childcare guidance to underserved communities, particularly targeting Africa’s 400 million adults without smartphone access. Glidance showcased a personal robot guide for blind individuals, priced at $14.99 with a $30 subscription, designed to help the 98% of blind people who cannot access traditional mobility aids. Predicteon demonstrated an AI system for personalized anesthesia management during surgery, aimed at reducing the $30 billion annual cost of anesthesia complications while improving patient outcomes. Finally, Revolutionize presented MAP, a smartphone-based malnutrition screening tool that uses AI to assess children’s growth through single-click photography, having already screened 200,000 children across India.

After deliberation, the judges selected Mama Mate as the winner, recognizing its potential to address maternal mortality in underserved communities through accessible, offline AI technology. All finalists were invited to join the Young AI Leaders Hubs initiative, highlighting the competition’s broader mission to foster AI innovation for humanitarian purposes.

Keypoints

**Major Discussion Points:**

– **AI for Global Health Solutions**: The competition showcased four AI-powered healthcare innovations addressing critical global challenges, including maternal health in Africa (MamaMate), mobility assistance for the blind (Glidance), surgical safety through anesthesia monitoring (Predicteon), and malnutrition screening in children (Revolutionize)

– **Accessibility and Affordability Criteria**: Judges consistently emphasized the importance of solutions being deployable in underserved communities, with specific focus on pricing, offline functionality, and adaptability to low-resource settings with limited connectivity

– **Real-world Implementation and Scalability**: Discussion centered on how these AI solutions could be practically deployed globally, with judges evaluating factors like market reach, technical feasibility, and potential for widespread adoption to address humanitarian challenges

– **Technology Integration and Safety**: Detailed examination of how AI is specifically utilized in each solution, including questions about data security, device interfaces, safety protocols, and the necessity of AI versus alternative approaches

– **Serving Marginalized Communities**: Strong emphasis throughout on solutions that could effectively reach and serve underserved populations, particularly in developing countries and rural areas with limited healthcare infrastructure

**Overall Purpose:**

This was the finale of an AI for Good Innovation Factory competition, where four finalist startups presented AI-powered solutions designed to address global humanitarian challenges and UN Sustainable Development Goals. The goal was to select a winner based on criteria including global impact potential, affordability, scalability, and ability to serve marginalized communities.

**Overall Tone:**

The discussion maintained a consistently positive, encouraging, and professional tone throughout. It began with excitement and anticipation, continued with focused technical questioning during presentations, and concluded with celebration and congratulations. The judges were supportive yet thorough in their evaluation, and the atmosphere remained uplifting even during the competitive deliberation process, emphasizing that all participants were already winners for their contributions to humanitarian AI solutions.

Speakers

– **Peter Vanham**: Editorial director of leadership at Fortune, moderator of the finale

– **Seizo Onoe**: Director of the standardization bureau of the ITU, judge

– **Ebtesam Almazrouei**: Chairperson of the UN AI for Good Impact Initiative, judge and jury chairperson

– **Werner Vogels**: CTO of Amazon, judge

– **Contestant 1**: Representative of Elevate AI Africa/Mama Mates (maternal healthcare solution)

– **Contestant 2**: Amos Miller, founder and CEO of Glidance (mobility solution for blind people)

– **Contestant 3**: Eva, representative of Predicteon (AI personalized anesthesia solution)

– **Contestant 4**: Romita, representative of MAP – Malnutrition Assessment and Action Plan (child nutrition screening)

– **LJ Rich**: Event coordinator/host for the award ceremony portion

Additional speakers:

None identified beyond the provided speakers names list.

# AI for Good Innovation Factory Competition Finale: Discussion Report

## Executive Summary

The AI for Good Innovation Factory competition finale brought together four pioneering startups to present AI solutions addressing critical global challenges. Moderated by Peter Vanham, Editorial Director of Leadership at Fortune, the event featured a distinguished panel of judges including Seizo Onoe (Director of the Standardisation Bureau of the ITU), Ebtesam Almazrouei (Chairperson of the UN AI for Good Impact Initiative), and Werner Vogels (CTO of Amazon).

The finale showcased four solutions spanning maternal healthcare, visual impairment assistance, surgical safety, and malnutrition assessment. Each presentation followed a structured format of two-minute pitches followed by three-minute question-and-answer sessions with the judges. Ultimately, Elevate AI Africa’s Mama Mate solution emerged victorious, though all finalists received recognition and were invited to join the AI for Good Impact Initiative.

## Competition Framework and Judging Criteria

The judges established clear evaluation criteria emphasizing humanitarian impact and practical implementation. Seizo Onoe emphasized the importance of solutions being “implementable, widely deployable, and adaptable.” Ebtesam Almazrouei stressed the critical importance of “AI solutions serving marginalised or underserved communities with affordable, scalable ideas.” Werner Vogels focused on “global challenge alignment and solving the hardest problems.”

The judges demonstrated consensus on prioritizing solutions that serve underserved communities while ensuring practical viability and affordability.

## Contestant Presentations

### Elevate AI Africa: Mama Mate – Maternal Healthcare Solution

Elevate AI Africa presented Mama Mate, an offline AI-powered device providing WHO-approved maternal and childcare guidance. The solution targets 91 million underserved mothers globally, with a goal of reaching 1% (910,000 mothers) in the first year. The device costs $25, features solar charging capability, and provides up to 36 months of WHO-guided care without internet connectivity.

The presenter emphasized that the solution addresses Africa’s 400 million adults without smartphone access. The AI chip represents the most expensive component at $10 of the total $25 cost. The solution was developed with input from African midwives and mothers and tested in the Chaga community in Kilimanjaro and Zulu community in South Africa.

When Werner Vogels questioned why a dedicated device would be more successful than SMS-based solutions, the contestant responded that “not everybody is literate to read an SMS and unfortunately connectivity is not everywhere.”

### Glidance: Personal Navigation for Visual Impairment

Amos Miller, founder and CEO of Glidance, presented a personal robot guide for blind individuals. The solution addresses the fact that 98% of blind people cannot access traditional mobility aids due to cost, training requirements, or availability constraints.

Glidance costs $1,499 for the hardware with a $30 monthly subscription, making it more accessible than traditional mobility aids. The device features an intuitive interface with steering wheels and voice commands, using “classic robotics with AI for semantic understanding.”

Judges questioned interface optimization, cognitive load reduction, and long-term safety protocols, particularly in complex urban environments.

### Predicteon: AI-Enhanced Surgical Safety

Eva from Predicteon presented an AI solution for personalized anaesthesia management during surgical procedures. The system addresses anaesthesia-related complications that cost healthcare systems approximately $30 billion annually. The company’s founder Pedro is an anesthesiologist who developed the solution.

Predicteon offers both dedicated hardware devices and software integration with existing medical equipment. The solution includes training capabilities for anaesthesiologists in remote settings through web-based software. According to Eva, hospitals can increase surgeries by 25%, unlocking up to $5 million in additional revenue.

Werner Vogels questioned the necessity of dedicated hardware devices, while Ebtesam Almazrouei asked about adaptation for rural communities with limited anaesthesiologist availability.

### Revolutionize: MAP – Malnutrition Assessment Technology

Romita, representing Revolutionize and drawing on her 20-year career as a public health leader, presented the MAP (Malnutrition Assessment and Action Plan) system. The solution uses smartphone-based photography to assess children’s growth and nutritional status, having already screened 200,000 children across India with a goal to reach 100 million by 2030.

The system employs AI-based height measurement and growth prediction using single-click smartphone images. Romita emphasized: “You shouldn’t measure us by the best of accuracy. You should see us as we are the zero intervention versus an intervention that MAP makes possible.”

The solution pursues an open-source algorithm approach while maintaining proprietary elements necessary for commercial viability.

## Judge Deliberation and Competition Outcome

During the judge deliberation period, with microphones turned off, Peter Vanham reflected on the broader context of the competition, mentioning a conversation with Meredith Whittaker about questioning whether AI is always necessary for solving problems.

Following deliberation, the judges selected Mama Mate as the competition winner. Ebtesam Almazrouei announced that all finalists would join the AI for Good Impact Initiative. The winner received a check, while all finalists received medals and certificates during a ceremonial presentation conducted by LJ Rich.

## Key Themes and Challenges

Throughout the presentations and questioning, several key themes emerged:

**Accessibility and Affordability**: All solutions prioritized reaching underserved populations through affordable pricing and accessible design.

**Offline Functionality**: Multiple solutions emphasized the ability to function without reliable internet connectivity, addressing infrastructure limitations in target communities.

**Integration vs. Independence**: Judges consistently questioned whether dedicated devices were necessary or if integration with existing systems would be more effective.

**Regulatory Considerations**: Healthcare solutions faced questions about regulatory approval processes and safety validation across different countries.

**Cultural Context**: Solutions demonstrated awareness of the need for community engagement and cultural adaptation in their development and deployment.

## Conclusion

The AI for Good Innovation Factory competition finale demonstrated the potential for artificial intelligence to address critical global challenges while highlighting the practical considerations necessary for humanitarian technology deployment. The competition’s emphasis on serving marginalized communities, ensuring affordability, and addressing genuine global challenges established a framework that prioritizes humanitarian impact.

The selection of Mama Mate as the winner, combined with ongoing support for all finalists, reinforced the collaborative nature of addressing global challenges through technological innovation. The competition provided valuable insights into the current state of AI for humanitarian applications, emphasizing the importance of community engagement, practical implementation, and sustainable deployment strategies.

Peter Vanham: Well, hello and welcome everyone to this amazing finale. I hope that this will give you a great insight into how indeed AI is being used to improve humanity and hopefully for a very long time to come. As you’ve just heard, my name is Peter Van Ham. I am editorial director of leadership at Fortune and I’m going to be joined by three amazing judges here on stage. I’ll present them to you in just a second and then we’ll hear after that right away from our four finalists. So without further ado, can I please invite our judges onto the stage? Excellent. I hope you’re all seated comfortably, that you’re ready for the next 25 minutes. It’s going to be a whirlwind. I’m going to briefly introduce you now, the judges. We have here Seizo Onoe, you’re the director of the standardization bureau of the ITU. We have, I have to come back to you because I had in my mind Werner Vogels Vogels of Amazon, the CTO. And then of course we have Ebtesam Almazrouei of, you are actually the chairperson. You will be chairing this jury, I suppose as well, but you are the chairperson of the UN AI for Good Impact Initiative. Welcome to all three of you. I’m going to ask you just one introductory question before we get started, which is very important for the people who are going to be joining us on stage right now. We all want to know how this finale is going to end, but you get to decide how it’s going to end. So for each of you, what are you looking forward to the most in this finale? What will be the defining criterion for you? Seizo, you start.

Seizo Onoe: Yeah, actually at this stage, it’s necessary to be the most profitable solution, but I hope it will be implementable and widely deployable, adaptable.

Peter Vanham: Excellent. The most widely deployable solution.

Seizo Onoe: Which saves the world by AI technology.

Peter Vanham: That’s exactly what we’re going to do. We’re going to save the world, not a small challenge. How about you?

Ebtesam Almazrouei: Basically, I’m looking at how they can translate AI into real world AI solution to serve marginalized or underserved communities with affordable, scalable ideas. Thank you.

Peter Vanham: And I know we’ve got a few contestants for that title, so that’s going to be a tough one I think. And Werner Vogels?

Werner Vogels: Global challenge alignment. Are these startups really solving some of the hardest problems out there.

Peter Vanham: Right. Okay. Well, we’ll see if they are going to be solving those UN Sustainable Development Goals or other global challenge that we’ve put ourselves as humanity. Whether they’re profitable, we’ll find out, but that’s not the most important criterion we know from you. And then now without further ado, it is time to hear from our finalists. Are you all ready? Yes, we are ready. Okay. So then I’m going to call onto the stage from number one to four. They’ll only have two minutes to present and we as a jury will only have three minutes to ask them questions. So this will be over before you know it. I want to ask to the stage first, our first contestant is Elevate AI Africa, please. And the floor is yours. Welcome.

Contestant 1: Thank you so much. So we asked ourselves a bold question. What does digital inclusivity and artificial intelligence mean in a world where 2.6 billion people remain completely offline? In Africa alone, over 400 million adults do not own a smartphone or access the internet. And while this gap is very deadly in maternal and healthcare, because for many mothers in underserved communities, the line of care is very limited. Every day, over 800 women and 7,000 newborn babies die from preventable childbirth and pregnancy complications. Every year, 5 million children under the age of five die from treatable conditions like dehydration and fever, just to name a few. Not because care does not exist, but because it does not reach the mothers who need it most on time. Mama Mates is our answer. Our innovation was built in silence, the kind of silence many women in underserved communities know too well. No doctor available on call. No guidance. No tools. Just fear. But this fear is not personal. It is systemic. And through our offline AI-powered solution, we seek to address and overcome this gap. Mama Mate delivers WHO guidelines, approved guidelines, guidance to mothers in underserved communities in order to help them bridge that line of care to avoid delay. From postnatal care tips to early infant nutrition guidance, all within a small pocket-sized gadget that is solar-chargeable and built to serve humanity. Thank you.

Peter Vanham: Thank you so much. And then we’re going to try to ask you one or two questions in the three minutes that we have. Abdisam, do you have a question? Yeah.

Ebtesam Almazrouei: How we can utilize Mama Mate in the areas where we have low connectivities?

Contestant 1: Thank you so much. So Mama Mate is built to be offline. It is integrated with a chip, an AI chip that is a neural decision processor chip, and it’s built for low literacy communities with a very audio interface-first approach. So even a mother who cannot speak and who cannot read will understand the communication because it’s also built in local languages. Thank you.

Peter Vanham: Thank you.

Seizo Onoe: Just one quick question about the, you are focusing on the African market first, African market first. Do you have any features specific to the African market or your solution can be applied for the global market?

Contestant 1: It is built for the global market, much as we’ve only tested it in two communities, one in the Chaga community in Kilimanjaro and one in a Zulu community in South Africa. But the idea is to go global because according to our research, the underserved market is 91 million mothers. And we seek to, in our first year, at least tap into 1%, which is 910,000 mothers of this market and truly enable mothers in underserved communities to flourish because no woman should die giving life and no child should die from a lack of adequate information and care. Thank you.

Peter Vanham: Werner Vogels?

Werner Vogels: How much will the device cost? Because affordability is a big issue.

Contestant 1: Thank you. So our most expensive aspect of the device is a $10 chip that we’ve built into it, but the device costs $25. It’s cheaper than a smartphone. It can be used offline. The outer parts of it can be 3D printed. It’s just the screen, the buttons, and the AI chip that costs more than what we’d want it to cost. But if we manage to have the right partners and scale it, which is part of our plan, we believe we can make this device much cheaper and more accessible to many mothers who truly would need it so that we do not have to lose 5 million children every year and have 800 mothers dying daily or 7,000 newborn babies.

Werner Vogels: No, definitely. Maternal health in Africa is a major issue and a major focus of many of the governments as well. This does mean they need to have an extra device. Thank you. If you take an approach like Chakandra Health, which would be based on just pure SMS because they have a dumb phone, why do you think an additional device will be successful?

Contestant 1: So I believe sometimes visual guidance and based on the cultural context, we built this device with the help of African midwives who attend to mothers and mothers themselves verified it and doctors verified it. So the approach is to ensure that it is a device that is durable, it’s solar chargeable. Not everybody is literate to read an SMS and unfortunately connectivity is not everywhere. So if we are able to give them a device that empowers them to up to 36 months of WHO guided care in the beginning of a child’s life, not only does it empower this child to have a better future health wise, but it also reduces on the mortality rates.

Peter Vanham: Excellent. Thank you so much and thank you for answering these very difficult questions in such a short amount of time. Thank you for joining us here. I think we’ve all enjoyed that. Thank you.

Contestant 1: Thank you. Can I make a comment? I didn’t get my slide shifter so I wasn’t able to shuffle through.

Peter Vanham: You did a fantastic job and we’ll see you very soon. Thank you so much and good luck. That was an amazing first contestant. Thank you so much. And now I want to immediately follow up with the second contestants, which is Glidance. Please join us on stage.

Contestant 2: Do you want to take my cue? Hi everyone. Are you here? Yes. Thank you. My name is Amos Miller, the founder and CEO of Glidance and I’m blind and I want you all to join me and close your eyes just for a moment and when I snap my fingers, you’ll open them again. But what if I never snap? You had sight for most of your life, you saw the smile of your loved one and now you wonder if you’ll ever feel independent again. You can open your eyes. This is the harsh reality for 300 million blind people today. To get back on your feet, you’re going to have to learn how to use a white cane or a guide dog. That can take five years, $100,000 and only 2% of blind people actually do that. Now, imagine if you had a personal robot that would be there, look after you, help you get around. You’d be back on your feet in no time and no, not a humanoid. I’m talking about a Glide, two wheels on the ground. You hold the handle and say, hey Glide, let’s go to Starbucks and Glide autonomously guides you there. A light and versatile device powered by a sensible wayfinding AI makes Glide easy to use and affordability is critical. Most blind people can afford a cell phone. Glide costs like a cell phone, making independent mobility accessible to millions. And they are lined out the door. We’ve collected almost a million dollars in pre-orders. Thousands have tested our devices and tell us This is as familiar as holding onto someone’s hand. With our team of engineers from iRobot, Google, Amazon and other places, Glidance is defining the category of personal robots. Look at the screen. If you lost your sight, tell me which one of those would you like to guide you around? Okay. Soon we will have thousands, then tens and hundreds of thousands of Glide in the wild where every step, this is two minutes. We made it already. Okay. Let’s better the lives of billions of people on this planet with the generational change that robots can bring to the world. Thank you.

Peter Vanham: Thank you also from us here in the jury. And Werner Vogels, I see you already raising your hands.

Werner Vogels: You mentioned an intuitive interface. Tell us a bit more about how do you interface with the device? So it’s an intuitive interface.

Contestant 2: Of course. So the most important thing is that Glide actually guides. You hold the handle, you walk forward. The wheels don’t even pull you. You push it forward. And as you begin to walk, Glide uses its cameras and begins to steer the way. The wheels steer. It can communicate with you with voice, tell you what’s going on. You can communicate with it to tell you where you need to go. You can press buttons. So very, very intuitive, very low cognitive load. And you just walk.

Peter Vanham: Excellent. Excellent. Sezu, do you have a question?

Seizo Onoe: Yeah. Just a pricing question. Which market do you see pricing? Which market?

Peter Vanham: Yeah. Which market do you see and what about the pricing?

Contestant 2: The pricing. Okay. So we talk about the market as the 2% who currently use canes and dogs and the 98% of people who are blind who are not able to get out of their homes. That’s the market. We are pricing it to be accessible to people, $14.99 for the device and a $30 subscription that you can make an Amazon order and get it.

Peter Vanham: Excellent. Thank you for that clarity.

Ebtesam Almazrouei: And I have a question if you don’t mind. How safe and secure to use this device?

Contestant 2: The device is using classic robotics that’s been around for 30 years at the bottom layer, if you like, for safety, to avoid obstacles, avoid cliffs, keep you from bumping into things. Cliffs are just drop-offs, not big cliffs. And then it uses AI for more, let’s consider it more a semantic stuff, like are you walking on the path or the lawn? So we can withstand some error there while keeping you safe.

Ebtesam Almazrouei: Thank you, sir.

Peter Vanham: Well, that’s excellent. It seems like you’ve already answered all of our questions, which is quite amazing. I’m just going to give the jury members one final chance because we do have a few seconds left. No further questions. That’s amazing. Absolutely. That means that you’ve made your case very, very clearly. Thank you so much.

Contestant 2: Thank you very much. And thank you for everyone. And to my team who’s on the video as well. Good luck.

Peter Vanham: Thank you so much. And then it’s already time for our third contestant. Our third contestant is Predicteon. And I would love to see if they can also join us on the stage. And when they do, please give them a very, very big applause for Predicteon.

Contestant 3: Hi, I’m Eva. At Prediction, we’re making surgery safer. 3.5 trillion euros are annually spent on surgeries. They represent 50% of hospital expenditure. Safe and efficient surgeries are critical. Without them, our hospital will run into financial trouble. Unfortunately, many hospitals are. Despite the high stakes, only 2% of AI FDA approved solutions focus on surgeries versus 80% on AI for diagnostics. And what do all surgeries have in common? Anesthesia. 60 million patients suffer anesthesia complications. During your surgery, your anesthesiologist, like our founder, Pedro, will spend hours mentally integrating thousands of data points, data tracking your heart, your lungs, your brain. Unfortunately, he is walking a fine line. Too little anesthesia, you would be in pain. Too much, you could suffer brain damage. And the cost? $30 billion annually on anesthesia complications. The solution, the Predixion PDA, AI personalized anesthesia. This is the medical device that we use in our hospital in Barcelona. During your surgery, it will integrate thousands of data points, and it will provide predictions to the anesthesiologists so they can adapt your level of anesthesia before you suffer complications. What it means to you as a patient, 80% less complications, 25% faster recovery. What it means to a hospital, they can increase surgeries by 25%, unlocking up to 5 million in additional revenue. The ask, 10 million euros to validate the results across hospitals around the world, including low and low middle income countries where there’s very limited access to anesthesiologists. For Predixion, anesthesia is just the beginning. Let’s make surgery safer together. Thank you to the Catalina AI factory, to the ITU, and all the organizers. Thank you.

Peter Vanham: Fantastic.

Werner Vogels: Why a dedicated device?

Contestant 3: Sorry?

Werner Vogels: Why a dedicated device?

Contestant 3: Very good question. This is a very, very big discussion. Why it’s not a software? Why is it a hardware? Why don’t we integrate our device or our software in other devices? We have two strategies, a hardware with software strategy, which is what you just saw. What it allows us is to go independently to a hospital. In parallel, we’re pursuing with medtech companies a software strategy, where we would put our software inside their devices. Thank you.

Peter Vanham: Excellent.

Ebtesam Almazrouei: How are we going to use this device in the rural communities?

Contestant 3: Thank you very much. I’ll start why can this even happen, yes? Before joining Predixion, I spent three years in Africa, in Rwanda, bringing oncology care in a rural setting. I was sure that this would be a very, very difficult hurdle. At the end, oncology requires you need to predict, you need to treat, et cetera. But for anesthesia, you only need an anesthesiologist. How can this device help? Number one, we can actually use the software to train anesthesiologists on remote. I can load different cases on my software. You don’t even need the device. It can be web-based. I can train anesthesiologists on how to see different cases. We can add the number of anesthesiologists. Number two, unfortunately, in many rural settings, we don’t have anesthesiologists. Sometimes there is a surgeon doing anesthesia. It’s not something that we would like to, but it’s a reality. Or even a nurse. We can simplify the solution for a different population, that they can use it in a very limited setting. Thank you.

Seizo Onoe: Okay. You need to use the data from hospital patients.

Contestant 3: Correct.

Seizo Onoe: So, is there any challenges or difficulties to get them data? How do you handle the data safely?

Contestant 3: Great question. We’re working, and we’re a spinoff of Hospital Clinic, one of the biggest hospitals in Barcelona, in Spain. And we follow very close protocols of clinical protocols on how to gather data, how to sign consent from our patients, and where and how to store the data, and who can access and when we can access that data. Thank you.

Seizo Onoe: No one receives it?

Contestant 3: No. It’s totally anonymized on top of that, obviously. Oh. The data is totally anonymized.

Werner Vogels: But the data that you do get, that you operate, is actually coming from other diagnostic devices in the hospital.

Contestant 3: Correct.

Werner Vogels: So, is there a standard interface to that that you can use?

Contestant 3: Yes. Yes. So, there is patient safety rules that allow everyone to connect to different devices. If I had my device and someone with an EEG device and an ECG device would not allow me to connect with their devices, they would actually be putting the patient at risk. So, you, patient safety regulations, allow all connections to different devices. It is true that our engineers have to code different drivers to connect to different devices. It’s not a hard task, but it’s definitely one of the tasks that we’re working on.

Peter Vanham: Excellent. Well, you’ve done it all. Thank you so much for answering the questions and for an amazing presentation. Thank you so much, Predicteon.

Contestant 3: Thank you.

Peter Vanham: Wow. Can you believe it? I told you it was going to be a whirlwind, right? So, it’s already time for our last contestant. I hope you’re all ready for it. Here we go with Revolutionize. Welcome.

Contestant 4: Good evening. This is the biggest stage I’ve ever been. I’m nervous. My name is Romita. As a child, my mother often told me, oil your hair regularly or it will turn red, like the begging kids. Years later, during my PhD in public health nutrition, I realized the heartbreaking fact. Their red hair is actually a sign of severe malnutrition. In a nation like India that feeds the world, no child should go to bed hungry, yet more than one third of the world’s malnourished children live in India. 73 million children have went unscreened, leading to stunting, wasting, and even deaths. That’s why we developed MAP, Malnutrition Assessment and Action Plan. With just one single click of an image, our AI-based technology can accurately measure a child’s height and predict growth irregularities based on WHO standards. All data is geotagged, tamper-proof, and encrypted. Every diet chart is personalized. MAP can also flag a lot of risk factors for a child. MAP is not just a technology. It’s a movement. We have already screened 200,000 children and are actively piloting with governments and schools across India. We have partnered and collaborated with NGOs for reach, academics for research, and AI labs for precision. I believe nutrition is not a United Nations problem. It’s everybody’s, and that’s why a part of our algorithm is open source. We believe collaboration is the key to solve this global challenge. Our goal? To reach 100 million children by 2030. In my 20-year-old career as a public health leader, I’ve already impacted more than a million lives, and to reach the next 100 million, I need your support. Together, let’s eradicate malnutrition from the face of earth. Thank you.

Peter Vanham: Thank you so much.

Werner Vogels: Tell me a bit exactly what the education material is that you provide.

Contestant 4: Absolutely. So, we are a population-level screening tool. We have found our sweet spot where the government, NGOs are already spending millions on nutritional programs. What they need is anthropometry, which means weighing machines, and studiometer that all these healthcare workers are supposed to carry and to reach these children in remote areas. What we are doing is you don’t need to carry scales anymore in 21st century. You just need to carry an application. These healthcare workers already have smartphones given by the government. We are strengthening the existing primary healthcare systems. We are built for robustness and for global scale. And to answer, we already have a new Gen-AI feature with just one single click of an image, you also get personalized nutrition plan. So, look at it as a point that where no intervention is happening, at least we are making that intervention.

Peter Vanham: Excellent. Abdisan?

Ebtesam Almazrouei: Yes. Is there any AI-based risk in utilizing the solution?

Contestant 4: So, a lot of regulators, we are doing a lot of validation study with EPICS bodies of India, with researchers, with academia, and they ask this question. Our question is that there is no particular framework. We help them understand that we are one of the successful healthcare, public health deployment tool for built for scale. You have to see us as, you shouldn’t measure us by the best of accuracy. You should see us as we are the zero intervention versus an intervention that MAP makes possible. Yes, there are risks and we have built safety by design. So all this data, we remove all the PIIs, we use masking, we use anonymization. So it has been thought through and it is a part of the design.

Ebtesam Almazrouei: Thank you.

Peter Vanham: Excellent. Seizo?

Seizo Onoe: Could you elaborate how do you use AI technology for your solution? For the product test. How do you use AI technology?

Contestant 4: Absolutely. So WHO already has a standardized format of growth monitoring for children of different age group, 0 to 2, 2 to 6. We are very active in the 2 to 6 range. They already have 12 ratios and with certain data, height, weight and age, you can calculate those ratios and tell whether a child is growing properly or not, is under, is stunted, is underweight, has malnutrition or not. So we have gamified those statistics. Traditionally, you have to calculate those Z-scores for standard deviation. So with AI, once we get accurate height, we collect weight manually, we collect age manually for each and every child, we give those ratios instantly and then we can map that child until that child becomes 10 years of age.

Peter Vanham: Excellent. Thank you so much. Wonderful job. And good luck. Thank you. Thank you. Okay. Wow. We already made it through these four finalists. I’m looking at you because as chairperson, I’m going to ask you now if you want to deliberate with the fellow judges for the next three minutes. It’s hard for the contestants, but it’s hard for you guys to find a winner, to designate a winner. But of course, they’re all winners. And please go ahead. I’ll ask now if your mics can be turned off so you can deliberate in peace. And I’ll just provide a few of my points of feedback now to the audience as well, because I think there’s a couple of things that I also wanted to share with you. Of course, we did go through this exercise in just about 20 minutes, which is a real short amount of time. And I want to highlight a few things. The first is that as you can imagine, the four contestants that came here today, they have gone through quite a lot of rounds to make it to this final. If you imagine how fantastic the solutions are that they offer, but also how great the needs are for humanity that are behind them, you can imagine that there’s actually underlying a lot more needs that need to be met and a lot more people that need to get that chance to do these kinds of competitions, to enter in these kinds of competitions. And I think that’s important to bear in mind. In a way, everyone here today is already a winner, but there’s many more companies that are trying to solve these needs, and everybody in the world should have the right and the ability to have their needs met. The second thing I’ll say is that you’ve just heard Cezo ask a question, how do you use AI? And of course, it’s a fair question to ask at a summit that’s all about AI. But I also want us to reflect on this and say, if it is not necessary to use AI, but you can still do something good, is there any good reason not to do that? I actually got reminded of that yesterday when I was talking to another person who was here on stage, Meredith Whittaker, and she challenged me on that very point. She said, Peter, you have to always ask that question. What problem are you trying to solve, and what’s the best way to solve that problem? And if AI is not the answer, that doesn’t mean it’s not a good problem to try and solve. So those are a few of the reflections that I wanted to share with you as I was participating in this exercise. And just one more time, before we get to the official end of this competition, I want to now ask all of you in the audience one final time to recognize the amazing work of the contestants that we’ve had here today. I hope you’ve all enjoyed listening to them. I hope they’ve all inspired you, and I hope you can all give them extra energy as they go and build their solutions. Once again, a big applause for our four finalists, please. And then I think that Werner Vogels is now going to catch the prize, because of course it is a competition, so we do have a prize to hand out for all the finalists, and of course for the winner. Werner Vogels, I want to make sure that we get that, so we can hand that out in a moment. But then I want to ask you, Abdussam, has the jury reached a conclusion? Who would you like to award as the winner of today’s competition?

Ebtesam Almazrouei: Yes, we reached a conclusion with my colleagues, and we are happy to, before we reveal the final winner, I would like to ask all our finalists to join the AI for Good Impact initiative, and they have a seat in the Young AI Leaders Hubs in their region. So welcome on board, everyone. And for the final winner, it will be MamaMate.

Peter Vanham: Congratulations. Congratulations. MamaMate, are you here? Congratulations. That was amazing. Really well done. Yeah, I won that. Thank you.

Ebtesam Almazrouei: This is for you. Congratulations. You did fantastic.

Peter Vanham: Thank you. Yeah. So we show the… Yeah, for him. This is for you. Oh, my God. I’m shaking. Well done. Thank you.

LJ Rich: Okay, everybody, congratulations, and we’re going to just get a little photo opportunity, and we can keep the music going for just a second longer, if that’s all right. There we go. Can we see everybody? Just hold the check a little bit down, please. Yeah, for me. Hold it a little bit down. And then congratulations. There you go. Okay. Thank you so much. You have your wonderful check. And all of our finalists, I would like to call all of our finalists back up to the stage. We want to give them a medal as well, because they have made it up to the final. So, finalists, please, will you join us on the stage? Come on up. Thank you. Our finalists, everybody. I’ll just place this down here a second. Let me come and… Congratulations. Well done. Okay. There we go. And we’ll get our photographs here. Let’s put this one on here. There we go. Congratulations. Congratulations. Wow. It’s… Hi there. Congratulations. I’m going to put a medal on you. There we go. I’m just putting it over your head now. Oh, my goodness. I know. You have a medal. It’s great. Yes. Ladies and gentlemen, our finalists and winners for the AI for Good Innovation Factory. And our incredible judges, you did a brilliant job as well. There we go. And also, Peter. Thank you. Thank you for running it. There we go. Congratulations. You may now relax as best you can. And that concludes our ceremony. Congratulations. Well done. Thank you. Thank you. Okay. Lovely. Well, that was pretty awesome, wasn’t it? Congratulations to all our finalists. Finalists, please can you head backstage? We have certificates for you as well. You were absolutely amazing. Wow. What a lovely way to start the final part of our evening. We do have a few more awards, as you well know, audience. And I’m going to be helping with this one.

Agreement points

Focus on serving underserved and marginalized communities

Speakers

– Ebtesam Almazrouei
– Contestant 1
– Contestant 4

Arguments

Focus on AI solutions serving marginalized or underserved communities with affordable, scalable ideas

Offline AI-powered device addressing maternal mortality in communities without internet access

AI-based height measurement and growth prediction using smartphone images

Summary

There is strong consensus that AI solutions should prioritize serving marginalized and underserved communities, with emphasis on affordability and scalability to maximize impact on those most in need.

Topics

Development | Human rights

Importance of affordability and accessibility in AI solutions

Speakers

– Seizo Onoe
– Ebtesam Almazrouei
– Contestant 1
– Contestant 2

Arguments

Solutions must be implementable, widely deployable, and adaptable

Focus on AI solutions serving marginalized or underserved communities with affordable, scalable ideas

Device costs $25, cheaper than smartphones, with solar charging capability

Pricing strategy of $14.99 device cost plus $30 subscription for accessibility

Summary

All speakers agree that AI solutions must be affordable and accessible to be truly impactful, with specific emphasis on pricing that underserved communities can afford.

Topics

Development | Economic

Addressing global health challenges through AI innovation

Speakers

– Werner Vogels
– Contestant 1
– Contestant 3
– Contestant 4

Arguments

Emphasis on global challenge alignment and solving the hardest problems

Offline AI-powered device addressing maternal mortality in communities without internet access

AI personalized anesthesia system to reduce complications during surgery

AI-based height measurement and growth prediction using smartphone images

Summary

There is consensus that AI should be used to tackle major global health challenges, from maternal mortality to surgical safety to malnutrition, addressing some of humanity’s most pressing problems.

Topics

Development | Human rights

Similar viewpoints

Both contestants emphasized the importance of working within existing healthcare systems and cultural contexts rather than replacing them, showing respect for local knowledge and infrastructure.

Speakers

– Contestant 1
– Contestant 4

Arguments

Built with cultural context using input from African midwives and mothers

Strengthening existing healthcare systems by replacing traditional weighing equipment

Topics

Development | Sociocultural

Both contestants prioritized safety and established protocols, with Contestant 2 using proven robotics technology and Contestant 3 following strict clinical protocols for patient safety.

Speakers

– Contestant 2
– Contestant 3

Arguments

Safety features using classic robotics with AI for semantic understanding

Data anonymization and clinical protocols for patient safety and privacy

Topics

Human rights | Infrastructure

Both contestants emphasized the importance of knowledge sharing and capacity building, whether through training programs or open source collaboration to maximize global impact.

Speakers

– Contestant 3
– Contestant 4

Arguments

Training capabilities for anesthesiologists in remote settings through web-based software

Open source algorithm approach to encourage global collaboration

Topics

Development | Sociocultural

Unexpected consensus

Recognition of all finalists regardless of competition outcome

Speakers

– Ebtesam Almazrouei
– Peter Vanham
– LJ Rich

Arguments

Recognition that all finalists deserve to join the AI for Good Impact Initiative

Recognition that many worthy solutions exist beyond the finalists

Ceremonial presentation of awards and medals to finalists and winner

Explanation

Despite being a competition with a single winner, there was unexpected consensus among organizers that all participants deserved recognition and ongoing support, suggesting a collaborative rather than purely competitive approach to AI innovation.

Topics

Development | Sociocultural

Questioning the necessity of AI for good solutions

Speakers

– Peter Vanham
– Seizo Onoe

Arguments

Emphasis on solving problems effectively regardless of whether AI is necessary

Solutions must be implementable, widely deployable, and adaptable

Explanation

Unexpectedly, at an AI-focused competition, there was consensus that the problem-solving effectiveness matters more than the use of AI technology itself, showing maturity in approaching technology as a tool rather than an end goal.

Topics

Development | Sociocultural

Overall assessment

Summary

The speakers demonstrated strong consensus around serving underserved communities, ensuring affordability and accessibility, addressing global health challenges, and prioritizing practical implementation over technological sophistication.

Consensus level

High level of consensus with significant implications for AI development priorities. The agreement suggests a mature understanding that AI should be human-centered, culturally sensitive, and focused on solving real-world problems for those most in need rather than pursuing technology for its own sake.

Different viewpoints

Hardware vs Software Implementation Strategy

Speakers

– Werner Vogels
– Contestant 3

Arguments

Why a dedicated device?

We have two strategies, a hardware with software strategy, which is what you just saw. What it allows us is to go independently to a hospital. In parallel, we’re pursuing with medtech companies a software strategy, where we would put our software inside their devices

Summary

Werner Vogels questioned the need for dedicated hardware devices, while Contestant 3 defended a dual approach using both hardware and software strategies for different market penetration methods

Topics

Infrastructure | Economic

Device vs SMS-based Solutions for Healthcare

Speakers

– Werner Vogels
– Contestant 1

Arguments

This does mean they need to have an extra device. If you take an approach like Chakandra Health, which would be based on just pure SMS because they have a dumb phone, why do you think an additional device will be successful?

Not everybody is literate to read an SMS and unfortunately connectivity is not everywhere. So if we are able to give them a device that empowers them to up to 36 months of WHO guided care

Summary

Werner Vogels questioned why an additional device was necessary when SMS-based solutions exist, while Contestant 1 argued that dedicated devices are better due to literacy issues and connectivity problems

Topics

Development | Infrastructure

Unexpected differences

Necessity of AI Technology for Problem Solving

Speakers

– Peter Vanham
– Seizo Onoe

Arguments

What problem are you trying to solve, and what’s the best way to solve that problem? And if AI is not the answer, that doesn’t mean it’s not a good problem to try and solve

Could you elaborate how do you use AI technology for your solution?

Explanation

While Seizo Onoe questioned contestants about their specific use of AI technology, Peter Vanham challenged the fundamental assumption that AI is necessary for good solutions, suggesting that effective problem-solving should take precedence over AI implementation

Topics

Development | Sociocultural

Overall assessment

Summary

The disagreements were primarily technical and methodological rather than fundamental philosophical differences. Main areas included implementation strategies (hardware vs software), delivery methods (dedicated devices vs existing infrastructure), and the role of AI in solutions

Disagreement level

Low to moderate disagreement level. The speakers generally shared common goals of creating impactful AI solutions for global challenges, but differed on technical approaches and implementation strategies. These disagreements were constructive and focused on optimizing solution effectiveness rather than opposing the core mission

Partial agreements

Similar viewpoints

Both contestants emphasized the importance of working within existing healthcare systems and cultural contexts rather than replacing them, showing respect for local knowledge and infrastructure.

Speakers

– Contestant 1
– Contestant 4

Arguments

Built with cultural context using input from African midwives and mothers

Strengthening existing healthcare systems by replacing traditional weighing equipment

Topics

Development | Sociocultural

Both contestants prioritized safety and established protocols, with Contestant 2 using proven robotics technology and Contestant 3 following strict clinical protocols for patient safety.

Speakers

– Contestant 2
– Contestant 3

Arguments

Safety features using classic robotics with AI for semantic understanding

Data anonymization and clinical protocols for patient safety and privacy

Topics

Human rights | Infrastructure

Both contestants emphasized the importance of knowledge sharing and capacity building, whether through training programs or open source collaboration to maximize global impact.

Speakers

– Contestant 3
– Contestant 4

Arguments

Training capabilities for anesthesiologists in remote settings through web-based software

Open source algorithm approach to encourage global collaboration

Topics

Development | Sociocultural

Key takeaways

MamaMate (Elevate AI Africa) won the AI for Good Innovation Factory competition for their offline AI-powered maternal healthcare device targeting underserved communities

All four finalists demonstrated solutions addressing critical global challenges: maternal healthcare, visual impairment assistance, surgical safety, and malnutrition assessment

Judges prioritized solutions that are implementable, widely deployable, serve marginalized communities affordably, and align with global challenges

All finalists were invited to join the AI for Good Impact Initiative and Young AI Leaders Hubs in their respective regions

The competition highlighted the importance of solving real-world problems effectively, whether or not AI is the optimal solution

Key success factors identified include affordability, scalability, cultural context integration, and addressing systemic gaps in healthcare access

Resolutions and action items

All finalists to join the AI for Good Impact Initiative as members

All finalists to participate in Young AI Leaders Hubs in their respective regions

MamaMate to receive winner’s prize and recognition for their maternal healthcare solution

Finalists to collect certificates backstage following the ceremony

Unresolved issues

Specific funding amounts or investment commitments for the solutions were not clearly established

Timeline for implementation and scaling of the winning and finalist solutions remains unclear

Regulatory approval processes and frameworks for healthcare AI solutions, particularly in developing countries, were acknowledged as ongoing challenges

Long-term sustainability and maintenance of devices in remote, resource-limited settings not fully addressed

Integration challenges with existing healthcare systems and infrastructure remain to be worked out

Suggested compromises

Predicteon’s dual strategy approach of both dedicated hardware devices and software integration with existing medical equipment to address deployment flexibility

Recognition that in resource-limited settings, non-ideal but practical solutions (such as surgeons performing anesthesia) may need AI support as an interim measure

Open source algorithm approach suggested by the malnutrition assessment team to encourage global collaboration while maintaining proprietary elements

What does digital inclusivity and artificial intelligence mean in a world where 2.6 billion people remain completely offline?

Speaker

Contestant 1 (Elevate AI Africa)

Reason

This comment reframes the entire AI discussion by highlighting a fundamental paradox – how can AI solutions truly be inclusive when billions lack basic digital access? It challenges the assumption that AI solutions automatically reach those who need them most and forces consideration of offline-first approaches.

Impact

This opening question set the tone for evaluating solutions based on real-world accessibility rather than just technological sophistication. It influenced the judges’ subsequent questions about connectivity, affordability, and deployment in underserved communities, making accessibility a central evaluation criterion throughout all presentations.

You shouldn’t measure us by the best of accuracy. You should see us as we are the zero intervention versus an intervention that MAP makes possible.

Speaker

Contestant 4 (Revolutionize)

Reason

This comment challenges the perfectionist mindset often applied to AI solutions, arguing that imperfect intervention is better than no intervention at all. It reframes the evaluation criteria from technical perfection to practical impact, which is particularly relevant for global health challenges.

Impact

This perspective shift influenced how the judges and audience viewed AI solutions in resource-constrained environments. It moved the discussion away from technical benchmarks toward pragmatic deployment considerations, emphasizing that good enough solutions that can be widely implemented may be more valuable than perfect solutions with limited reach.

What problem are you trying to solve, and what’s the best way to solve that problem? And if AI is not the answer, that doesn’t mean it’s not a good problem to try and solve.

Speaker

Peter Vanham (referencing Meredith Whittaker)

Reason

This meta-commentary challenges the fundamental premise of an AI-focused competition by questioning whether AI is always the optimal solution. It introduces critical thinking about technology selection and problem-solving approaches, suggesting that the tool should serve the problem, not vice versa.

Impact

This reflection provided crucial context for evaluating the presentations, encouraging the audience to think beyond the AI hype and focus on problem-solving effectiveness. It validated solutions that used AI appropriately rather than innovatively, and helped frame the competition as being about impact rather than technological novelty.

Our innovation was built in silence, the kind of silence many women in underserved communities know too well. No doctor available on call. No guidance. No tools. Just fear. But this fear is not personal. It is systemic.

Speaker

Contestant 1 (Elevate AI Africa)

Reason

This comment powerfully humanizes the technical solution by connecting it to lived experiences and systemic inequalities. It transforms the discussion from a technical pitch to a social justice narrative, emphasizing that technology solutions must address systemic rather than individual problems.

Impact

This framing elevated the emotional stakes of the competition and influenced how subsequent presenters positioned their solutions. It established that winning solutions needed to demonstrate deep understanding of the communities they serve, not just technical capabilities. The judges’ questions increasingly focused on community engagement and systemic impact.

Global challenge alignment. Are these startups really solving some of the hardest problems out there.

Speaker

Werner Vogels

Reason

This criterion cuts through potential technological showmanship to focus on genuine global impact. It challenges presenters to demonstrate that their solutions address fundamental human needs rather than creating solutions in search of problems.

Impact

This evaluation framework shaped how all contestants positioned their solutions, forcing them to explicitly connect their work to major global challenges like maternal mortality, disability access, surgical safety, and malnutrition. It elevated the discussion from startup pitches to humanitarian impact assessment.

Overall assessment

These key comments fundamentally shaped the discussion by establishing a framework that prioritized human impact over technological sophistication. The conversation evolved from a typical startup pitch format to a deeper examination of how AI can address systemic inequalities and global challenges. The winning solution (MamaMate) succeeded not because of its AI innovation, but because it most effectively addressed the criteria established by these pivotal comments: serving underserved communities, working offline, addressing systemic problems, and providing practical intervention where none existed before. The discussion demonstrated that in the context of AI for good, the most thought-provoking insights came from questioning assumptions about technology deployment, accessibility, and the relationship between technical perfection and real-world impact.

How can the AI chip in Mama Mate be made more affordable through scaling and partnerships?

Speaker

Werner Vogels

Explanation

The $10 AI chip represents the most expensive component of the $25 device, and affordability is critical for reaching underserved communities in Africa and globally

Why would an additional device like Mama Mate be more successful than SMS-based solutions like Chakandra Health for maternal care?

Speaker

Werner Vogels

Explanation

This addresses the fundamental question of device strategy and whether introducing new hardware is more effective than leveraging existing infrastructure like basic phones

What are the specific features that make the solution adaptable to different African markets versus global markets?

Speaker

Seizo Onoe

Explanation

Understanding market-specific adaptations is crucial for scalability and determining whether the solution can be effectively deployed across diverse cultural and linguistic contexts

How can the Glidance device interface be further optimized to reduce cognitive load for blind users?

Speaker

Werner Vogels

Explanation

While described as intuitive, the specific mechanisms of voice communication, button interfaces, and steering feedback need further exploration for optimal user experience

What are the long-term safety protocols and fail-safes for the Glidance navigation system in complex urban environments?

Speaker

Ebtesam Almazrouei

Explanation

Safety and security are paramount for a device guiding vulnerable users through potentially dangerous environments, requiring comprehensive testing and validation

How can Predicteon’s anesthesia AI be adapted for use in rural communities with limited anesthesiologist availability?

Speaker

Ebtesam Almazrouei

Explanation

The solution needs to address the reality that many rural settings lack trained anesthesiologists, requiring adaptation for use by surgeons or nurses with limited anesthesia training

What are the standardization challenges for connecting Predicteon’s device to different hospital equipment across various manufacturers?

Speaker

Werner Vogels

Explanation

While patient safety regulations allow connections, the practical implementation of different drivers for various devices presents ongoing technical and standardization challenges

How can data privacy and security be maintained when handling sensitive patient anesthesia data across different healthcare systems?

Speaker

Seizo Onoe

Explanation

Managing anonymized patient data while ensuring clinical effectiveness requires robust protocols that may vary across different healthcare regulatory environments

What regulatory frameworks need to be developed for AI-based healthcare screening tools like MAP in different countries?

Speaker

Ebtesam Almazrouei

Explanation

The lack of specific regulatory frameworks for AI healthcare tools presents challenges for global deployment and requires development of appropriate safety and efficacy standards

How can the accuracy limitations of image-based anthropometry be balanced against the need for population-level screening in resource-limited settings?

Speaker

Seizo Onoe

Explanation

Understanding the trade-offs between perfect accuracy and practical implementation is crucial for validating AI-based screening tools in public health contexts