From Drones to AI: PixForce's Journey in Transforming Inspections with Daniel Mora

Summary

Discover how PixForce transforms industrial inspections with AI, turning visual data into actionable insights that enhance safety and efficiency. Founder Daniel Mora shares how PixForce evolved from manual tree counting to serving giants like Petrobras with AI-powered cameras that detect issues in real time. Learn about the challenges of scaling AI in complex industries and how PixForce's PlataPix platform simplifies deployment. This episode reveals the steps to demonstrate clear ROI and leverage AI for competitive advantage, making it essential for engineers, CTOs, and innovation executives looking to revolutionize asset management.

Takeaways

Validation over novelty
Scaling complexity
Context-aware AI
Adoption hurdles
Platform universality
Market maturity
Strategic growth

Soundbites

"We moved from manual processes to using deep learning for efficiency."
"Our next step is US expansion, with a presence in Houston."
"Large language models are making us more efficient in programming and training."

Chapters

00:02 Introduction - Evan introduces the show and Daniel Moura.
00:33 PixForce's Mission - Applying computer vision to industry.
01:27 Founding Story  - How PixForce began.
03:27 AI Transformation - AI's impact on PixForce.
06:23 Current Focus (6:23) - Solving industry problems.
18:10 Growth Plans (18:10) - US expansion and goals.
21:41 AI Trends (21:41) - Exciting developments in AI.
25:14 Future Goals (25:14) - Vision for success next year.

Video

https://youtu.be/nhuebPdqeso

Transcript

SPEAKER_00 0:20

Welcome to AI Speed, the show where AI-powered companies talk about what actually works in the market right now. Business doesn't move at internet speed anymore, it moves at AI speed, and the people who figure out how to turn models into money will own the next decade. I'm Evan J. Cholfin, founder of Luxhammer and growth partner to high-performing brands. Today, I'm thrilled to be joined by Daniel Mora, founder and CEO of PixForce. Daniel is building one of the leading AI companies, applying computer vision to real-world industrial challenges from safety monitoring to infrastructure inspection. PixForce is helping companies transform how they analyze images and video, turning massive amounts of visual data into actionable insights that improve safety, efficiency, and productivity. Daniel, thanks for being here.

SPEAKER_01 1:19

Thanks a lot, Delvin. It's a pleasure. And by the way, great introduction. I like it.

SPEAKER_00 1:25

We'll write your bio for you.

SPEAKER_01 1:28

Yeah, that's awesome. We did a better job than we'll do, so yeah.

SPEAKER_00 1:33

Yeah, AI helped me with that one. Good. I love it. Um so uh I understand PixForce is tackling a very complex problem, uh, automating visual inspections and interpreting images at scale. Was the pivotal uh moment or industry pain point that led you to identify this opportunity and start PEXForce?

SPEAKER_01 2:05

Yeah, that's that's great. Uh we started 10 years ago, so I came from uh the environmental sector. I was an environmental consultant, and I saw a very big opportunity to analyze images generated from drones, basically, out in the field. And my co-founder, Renato, he was coming from the mining sector, and he saw the same thing, you know, we had more and more cameras recording activities in the field, more drones. But the challenge was okay, who's gonna analyze all this footage? If you record a scene for eight hours, you're gonna have to have somebody going through that for eight hours. You know, there's nothing you can do about it. So at that point, we we're not talking about AI as we were talking today, but uh we were talking about software, you know, uh different types of strategies to analyze and extract information from this uh this footage. And that that's how PixForce started. And then everybody that's involved in in this type of business or any business like a startup, you have to be ready to adapt and and change and move, and we change a lot throughout the years, but our main thing was analyzing images, and we were still doing the same.

SPEAKER_00 3:22

Yeah, and so how has implementing AI changed your uh process and your company?

SPEAKER_01 3:30

Yeah, it changed a lot because in the beginning we were using some traditional software and everything was very manual. So our first client, just for as an example, since the beginning, we decided we had to have a very strong go-to-market because uh it doesn't matter that you have a great technology if you don't have clients that are willing to pay, it doesn't matter, right? So we we try to decide which sector we would start, and we decided to start in the pulp and paper industry because at that time we saw that they really had this pain of having to analyze like huge areas because these guys are like planting trees, like eucalyptus or pine trees, and they have to cut these trees in seven, ten years, but uh they didn't really know how many uh trees they had because they they could have had like a fire or a pest. So they had to do this inventory once in a while and see how many trees they had. And that they had to go out in the field and counting trees. You know, it was very manual, very, very tough work. At that point, 2016, they started buying these drones and surveying these areas, but they still had the problem of counting the trees. So now instead of going out in the field, they had to count in the office, you know. So they hired Peaks Force to run different types of software and to figure out a way to count these trees faster. And in the beginning, we were trying and trying, and we were doing a lot of manual work, you know, counting trees. So we had like 30 people counting trees, like massive areas. And then in one year, we decided we had to move to something more efficient, so we start using deep learning. So it's it's not uh Gen AI, but uh it's it's uh strategy for computer vision that's still used, you know. It was developed by companies such as Google and and other companies, so it's it's it's a very good way to teach the the machine what is a tree and what's not a tree or what what is a bush. So with that mass of data, so we had like millions of trees that we had count count manually, we were able to train a model, a computer vision model, and then we had great results. So after that, we we started applying this to other sectors such as electric, mining, oil and gas, and we evolved a lot since then.

SPEAKER_00 5:57

Yeah, definitely. And so what's the core problem you're solving for companies today, especially those operating in industries like energy, mining, or infrastructure?

SPEAKER_01 6:09

Yeah, that's good. We still do some counting, but that's not our main thing. Our main thing now is uh it's uh inspections in the field. And when you think about any industry, you have hundreds of people out there inspecting the work, you know. So they're seeing if people are working properly, if they are safe, if there's any machine that's broken. Similar to when you do a renovation or house, right? You you hire a contractor, but once in a while you want to go there and see if they're working correctly, you know. So all these industries are doing the same, but they have many people that are doing inspections, but uh they can't see everything, you know. And even if they see everything, just one person, you know, they can't be everywhere. So our idea is to use cameras that are already installed in factories or uh in trucks, or even like in a tripod, in a cell phone, or drones, you know. So any image that we can acquire from the field, we can basically see if you have problems with people and problems with assets, such as equipment. So that's basically what we're doing for our clients.

SPEAKER_00 7:20

That's great. And so where are you focusing most of your energy right now as founder and CEO?

SPEAKER_01 7:27

Uh as a CEO, I have to be involved in everything, you know, and we are we have uh 100 people now, so so we're not that big at this point. We had uh very interesting growth in the past three years, but we're still 100 people, so I have to be involved in fundraising, finance, the development, the product, and of course sales. You know, if I if I have to break down, I would say that sales is probably like 40%. Um another thing that I have to do a lot is always hiring. The hiring process, that's very important because every every everybody that we hire, uh I have to either talk to the person or at least oversee what's happening because I want to make sure that we're bringing bringing the right people. In the end of the day, uh we have great technology, great clients, but our biggest asset is our people, you know. And I don't say that just to sound nice, you know, it's the reality, you know. If you don't have the right people, it doesn't matter, you know, it's you're gonna fail. So I put a lot of energy also trying to recruit, you know, the right people.

SPEAKER_00 8:31

Absolutely. So who do you serve best? What kind of organization gets the most value from your AI solutions?

SPEAKER_01 8:38

Yeah, the in general, large companies. You know, the larger the company, the larger the risk, the larger the complexity or the size of the plants that have to be overseen. So we work for companies such as like very large oil and gas companies like pipeline companies. One of our clients is Petrobras, the largest oil and gas company in in South America. We work for companies in the mining sector like Anglo-American, ArcelorMittal, one of the largest uh smelters in the world. So uh a lot of electric utilities. We have offices in Brazil, Europe, and US, for example, in Portugal. We work for the national energy company there, EDP. So all these companies that have a lot of complexities and a lot of uh stuff to inspect.

SPEAKER_00 9:27

Absolutely. So looking at some of the challenges that you've been through, what do you think has been your biggest challenge growing Pixforce this past year, especially in convincing traditional industries to adopt AI and computer vision solutions?

SPEAKER_01 9:44

Yeah, that that's an interesting question because in the past, when we started, it was very hard to convince the clients that AI was something useful or something that they could rely on. So we had to show uh use cases and convince them that I have been doing the work the way you're doing for 100 years, but maybe you should change now because there's a better way. Now, everything changed because everybody wants to use AI, right? So some companies they come to us and say, we have to use AI, you know, give you something, you know. So it it's it's good in this point, but on the other hand, with like social media and all of that, some clients they see what we are doing, they say, okay, we are doing something similar. Or it doesn't sound that difficult because I saw this Instagram post that people were detecting personal protective equipment in people out in the field. So maybe what you're doing is not very unique. So that that's another problem because people think that a computer vision is straightforward and it's not. You know, it's very hard to scale a product. Maybe you can do for one camera, but if you want to do real-time analysis for a thousand cameras in five different plants, sending alerts in real time and uh having all of this with high accuracy, that's very hard. So one of the problems I see now is that people think that it's very easy. You know, it's like writing an email on Chat GPT. You know, that that's fantastic, but what we're doing is much harder, you know. And one thing that I always say is that there's a reason why, in general, the security way to check if you are a human being is through image analysis, uh capture, you know, right? So we have to select hats, select uh animals or a bicycle. Because for a computer, it's not straightforward to do image analysis, you know. It's much easier to do like a math equation than to do image analysis. So we are still uh, I would say, in the infancy of computer vision compared to other applications of AI. And this is hard to explain some some clients, you know. So that that's that's a challenge for us.

SPEAKER_00 12:01

Yeah, what do you do to address uh some of those challenges?

SPEAKER_01 12:05

Yeah, one of the things we have been developed uh in the past uh three years is a platform. It's called Platopix. So it's a platform that we can connect cameras very easily. So it doesn't matter the brand. You can connect a camera in our office, you know. It's it's very easy. You connect to our platform, and we have like uh a testing for for two weeks. We can connect two or three cameras. We offer there to our that to our big clients and say, okay, let me show you what we can do. We don't charge for that. We send the data. So that's that's a very good way for them to see the capacity and the accuracy of our platform. And other ways to show that is true. Use use cases, as I said in the beginning. You know, we we have clients in the same sector that are doing similar applications with our AI, and we show them, and maybe we can give them as a reference, they can talk to them, and that helps, you know. But uh it doesn't matter. These these big clients, it takes time to convince them, you know, and we know that when we are prepared for that.

SPEAKER_00 13:09

Yeah, definitely. I I know sometimes, you know, working with major corporations or organizations, it can be a slow process. Yeah. Even just with procurement after they say yes. Yeah.

SPEAKER_01 13:21

Yeah, there's a client, for example, we signed a contract, and procurement after two months said, okay, we did something wrong here. We had to have this contract passing through our headquarters and another country, we have to do it again, you know. So we they have to basically uh cancel our country and start all over, no? And what can you say, you know, you're not gonna complain. Say, okay, yes, if that's your process, no problem. But but it's it's frustrating, right? Yeah, definitely.

SPEAKER_00 13:52

So what kind of makes it difficult to get through the hype around AI because there's a lot of hype, obviously. Yes. And demonstrate that real ROI to the enterprise buyers.

SPEAKER_01 14:03

Yes, that that's that's uh very important, and uh we always try to show this ROI and and show that uh this is not a hype, you know, this is something that's very real and it's gonna provide a very great value for them. Right now, it's not for the future, you know. I think the best way is to do real testing, you know. I talked about the the way we have these uh these trial systems, and some sometimes for the drones, for example, we do pilots. So the clients say, okay, come here to our plant. I was talking like two days ago to a US company, like they are a mining company. And I said, okay, we can send a crew out there and you just pay for our costs and we're gonna show you our capability. So they have to see with their eyes, you know, and of course we are we're gonna generate data and results and try to calculate this ROI beforehand so they know their savings and what's the benefit for them, you know. Sometimes it's hard, for example, for health and safety, it's hard. You know, what's the value of one person's life, you know? How much does it cost an accident or a person that dies? Of course, you can use other ways to calculate that, like how much money you're saving with insurance, or if you have less lawsuits, or you know, there are other ways you can calculate, but sometimes not straightforward. And we need a client to help. We we we always need somebody inside that really wants uh to pursue this, so they help us uh estimating this ROI. Because sometimes we don't have the data on the inside.

SPEAKER_00 15:48

Yeah, absolutely.

SPEAKER_01 15:50

We have to engage them.

SPEAKER_00 15:51

Yeah, having the internal champion to kind of help guide you in the process as well makes a big difference. So along those lines, what are your main growth targets for 2026 for the company? I'm curious, are you looking to expand beyond uh your current territory within Brazil?

SPEAKER_01 16:09

Yes, yes. Uh our our next step now is uh U.S. expansion. We have a presence in Houston. We we just we we had a Delaware company and we opened a company in in Houston. So and the reason is because of all the oil and gas companies are there, you know, and we have very strong connections with all these major oil companies, so it makes sense to be there. Uh Houston is a very interesting place because there's this concentration of a single industry, you know. So we we we like that a lot. So so our goal this year, uh 2016, next year is to expand uh in the US. We have uh one person that we are hiring now in Houston. Uh of course I I go there all the time. I'm gonna be in Houston two weeks. My co-founder is also going there all the time. And uh we have also investors in the US. At this time we just had uh angel investors, we didn't have any big uh funds, but these uh investors are also helping with their network and with their connections. So we are very confident that uh in uh 26 we're gonna have our first uh three, four clients, and then next year we're gonna have our Series A. So we're gonna raise more money so we can do like an even better job expanding to the US.

SPEAKER_00 17:34

Yeah, that's exciting. So it sounds like you already will have some warm leads going in, uh, which is also good. Are you also planning to have sales to expand that pipeline beyond those initial leads?

SPEAKER_01 17:51

Yes, we are in talks with like a lot of uh companies, and companies uh I'm talking a lot about oil and gas, but we are talking to other sectors like the food industry, you know. So we have a major client that uh we are in the negotiations now uh in the US. So there are other industries that uh and that came from one of our investors, you know, he's very well connected with this uh this specific company. Right now we're gonna try to use our network. In Brazil, we are doing a lot of uh digital marketing, so we have a pre-sales team that do like uh code calls and all of that. Uh, of course, we're using AI also to approach our clients, you know, through social media. But uh we are trying to put a lot of effort in Brazil now with these strategies. And in the US, it's gonna be more like uh real connections we have, you know. So for this year, we're gonna expand more organically. And next year with our Series A, we want to have like this real exponential growth in the US. Yeah, that's great.

SPEAKER_00 18:55

So, what trends are you seeing in AI and computer vision that excite you most right now?

SPEAKER_01 19:01

Yeah, there are there are several, you know. I was uh describing what these large language models can't do, but they can do a lot, you know. So I'm not saying that they are not useful. We are very excited, starting with the programming part, you know, we can be more efficient, you know, programming our platforms, and that's giving us a lot of ad because we're being more and more efficient. And also the training part, because when you work with computer vision, you need a lot of images, and you have to annotate all these images, you know, we have to explain the machine okay, this is right equipment, this is broken, this is a missing part, and everything needs to be annotated. It's similar to the way we teach like an intern that started our company. We have to go through a lot of examples, and we have to do that for the machine. Uh but the way we used to do was all manual. You remember the beginning I was talking about the trees that we were clicking, you know, counting trees. Now with GNAI, we can do this work much faster. No, so GNAI can help us annotating this image. Sometimes gonna make some mistakes, and then we we come back and we we improve that. Uh but uh we can definitely be much more efficient doing this annotation process. So that's uh that's changing our business. It's going much faster. Also, I see that with uh Genai we can uh explore areas that we were not exploring before, like this agents. So with agents, we uh we are generating a lot of data from the images, right? So uh in the past, we used to have this database with all this data, like a data lake that you can use that for many uh applications, but we used to need like somebody looking the data or seeing the alerts. But now we can have an agent that's seeing everything and it's adding other information for health and safety, for example. We're seeing that we always have a problem, a person in risk in this specific area. Maybe if they ask us, we don't know the context about the factory, we don't know what what's going on, you know. But if you have an agent that has access to the plan drawings, to the health and safety plans, to to the health and safety meetings, the minutes of the meetings, and they know the name of the different contractors, the hours they work. So this agent starts understanding all the context and giving much more refined information for our clients. So we are very excited to merge this data from AI and this data from their offices, the back office, you know. And that's going to be very powerful to have a super agent that understands much more than anybody could understand about their own plants.

SPEAKER_00 21:57

Absolutely. So if we were to have this conversation. Again in 12 months, what would make you feel it's been a successful year for Pixforce?

SPEAKER_01 22:05

Yeah, definitely this US expansion. We finally have like a few products that are very scalable. We are deep tech, so we spent a lot of time our first years. We were developing our platforms, our AI, but now we have something that's uh really scalable. So I would be very happy if next year we talk and you know we have a handful of clients in the US, big clients, you know, and as we discussed, it's not easy, you know, it's not like I'm gonna pick up the phone and they're gonna hire me yet tomorrow. No, it's it's hard. So if we have a handful of clients in the US, I would be very happy and we're gonna be very well prepared for our plan for next year.

SPEAKER_00 22:45

Well, that's exciting. That's it for today's episode of AI Speed. A huge thank you to Daniel Mora for sharing his invaluable insights into how Pixforce is helping industries transform visual data into actionable intelligence through AI and computer vision. And if you're building or leading an AI native company or a service business that uses AI under the hood, and you care about revenue, adoption, and market share, make sure to subscribe to AI Speed. Learn how the best AI operators ship faster, sell smarter, and stay ahead. Thanks for listening. Until next time, keep building, keep selling, and keep moving at AI speed.