AI Minds #049 | Abdulrahman Jamjoom, Co-Founder and CEO at Arini AI
About this episode
Abdulrahman Jamjoom, Co-Founder and CEO at Arini AI, Arini is the complete AI solution for patient communication in dental.
Listen to the episode on Spotify, Apple Podcast, Podcast addicts, Castbox. You can also watch this episode on YouTube.
In this episode of AI Minds, Abdulrahman Jamjoom, co-founder and CEO of Arini AI, shares his inspiring journey—from his early days in Saudi Arabia to pursuing education in Jordan, and his entrepreneurial path in the AI space. After a transformative experience at Harvard and his time at Threads, a San Francisco-based startup, Abdul reflects on the inception and evolution of Arini AI.
He dives into how Arini leverages artificial intelligence to revolutionize patient communication management for dental practices, primarily in the U.S. and Canada, and the impact it's having on streamlining operations and improving customer experience in the dental industry.
Here are some highlights from this great episode:
Journey into AI: Abdul began his story discussing his journey from Saudi Arabia to the United States, highlighting his passion for startup culture which was ignited during his time at Harvard. He spoke about his initial steps in the tech world at Threads, emphasizing how this experience refined his product intuition and AI application skills.
Founding Arini AI: Delving deeper into his entrepreneurial path, Abdul shared the inception story of Arini AI. Born from firsthand observations of operational chaos at healthcare facilities, Arini AI was created to streamline front desk operations through AI-driven communication automation. Abdul’s vivid storytelling painted a clear picture of the immediate needs in healthcare communication, which his company aims to alleviate.
AI Solutions and Challenges: Discussing the technical side, Abdul elaborated on Arini AI’s capabilities in handling complex patient communications across multiple channels. He also touched on the challenges of ensuring high-quality AI interactions, adapting to different practice management systems, and the importance of scalability and reliability in AI solutions.
The Future of AI in Healthcare: The episode concluded with a look toward the future implications of AI in healthcare settings, particularly in how AI can drastically reduce the need for extensive human intervention in administrative tasks, thereby allowing healthcare providers to focus more on patient care.
Fun Fact: Abdul tried really hard not to graduate from Harvard because he was so eager to start his own company. His passion for entrepreneurship was so strong that he saw his educational journey as a ticking clock towards starting his own business.
Show Notes:
00:00 Started in Saudi, now thriving in startup life.
05:12 Rami and I wanted to start company.
08:12 Staffing shortages worsen front office operations.
12:10 Arini manages communication, improving efficiency and ROI.
14:58 Improving AI interaction for confident user engagement.
17:08 Focus: Fast, efficient booking matters more than realism.
20:18 Scheduling software usage varies across different groups.
23:29 Assistance provided despite booking unavailability, peak hours.
More Quotes from Abdul:
Transcript:
Demetrios:
Welcome back to the AI Minds Podcast is a podcast where we explore the companies of tomorrow being built AI First. I'm your host, Demetrios. And this episode, like every episode, is brought to you by Deepgram. The number one speech to text and text to speech API on the Internet today. Trusted by the world's top enterprises, conversational AI leaders and startups. Some of them you may have even heard of like Spotify, Twilio, NASA and Citibank. In this episode, I am very thankful to be joined by the CEO and co founder of Arini AI Abdul. How you doing today, man?
Abdulrahman Jamjoom:
I'm doing well. How are you? Very excited to be here.
Demetrios:
Likewise. Well, honestly, I love your story. You started in Saudi Arabia, you got educated in Jordan, and then you took a turn for the us you decided you wanted to live that startup life after a stint at Harvard. And one thing that you told me is you tried as hard as you could not to graduate Harvard because you wanted to create a startup before you got done with your education, because you had it in your blood and now you are living out your dreams. But before we talk about what you're doing with Arini AI, I would love to hear about your time in startup land in San Francisco because I think it's fascinating when you were working at Threads.
Abdulrahman Jamjoom:
Yeah, absolutely. I mean, after graduating, I joined Threads. I was working as a full stack engineer there and really focused on search in the beginning. And the Threads team is absolutely incredible. Like, really the reason I joined is that I wanted to learn from the best product minds, engineers, designers. And I'm very grateful that I got to do that because I think it really shaped a lot of my product thinking, product intuition. So, yeah, it was a great time there. Spent a lot of time on search and then transitioned more into applied AI.
Abdulrahman Jamjoom:
So that's around the time when ChatGPT started coming out. The APIs were out. So everyone was thinking, okay, how do I take this tool and enhance the product experience? just so for some context, Threads is a communication platform. So think Slack for companies, an async and sync tool. So we're doing a lot of things like prioritizing notifications, summarizing conversations and so on.
Demetrios:
Yeah, yeah, it is funny. I should have caveated Threads with. It's not the threads that people are probably thinking about today because of the whole Twitter challenger of Zuckerberg's threads, but I knew about Threads from back in the day because I loved listening to podcasts and just ways that this guy who is very big On Async communication talks. And I think he's the CEO of Levels Health or. Level. Yeah, Levels Health or something like that. They talk about it all the time.
Abdulrahman Jamjoom:
Yep, absolutely. I remember the feedback he used to send. Worked a lot, a lot on this. Of the stuff that he was suggesting and we were discussing at the time.
Demetrios:
Yeah, well, so you spent time doing search. I bet those search chops really came in handy when Rag started to become the normal design pattern.
Abdulrahman Jamjoom:
Honestly, it was probably the best investment. One of the best investments that I got was like spending a lot of time figuring out search. And also so actually, other than search, I spent a lot of time on triaging flows because, like, when you think of a notification system, notifications are coming, you want to prioritize them. And then once we get to like discussing Irini a bit more in depth, like triaging, notification becomes so critical, because you have to tell people how. How things are going, like, what are the agents doing? So that ended up being super, super, super valuable for me.
Demetrios:
Fascinating. So in the context of Threads, you were thinking about notifications, like what should we show someone as far as if they got mentioned or what is the most important piece there and how do.
Abdulrahman Jamjoom:
You allow them to triage it as fast and efficiently as possible?
Demetrios:
Wow.
Abdulrahman Jamjoom:
So that's the critical part is like, if they have like 200 notifications, how can they like triage them in 15 minutes as opposed to like three hours? How do you populate the important things up top and tell them, hey, this is critical, you need to check it out as opposed to just like weeding out the signal from the noise is so critical and I think so many companies actually miss that because it is a very hard problem to solve. Proper notification triaging.
Demetrios:
Yeah. Knowing what is important versus what is just. You got tagged in random on a meme.
Abdulrahman Jamjoom:
Yeah, absolutely.
Demetrios:
So you spend some time at Threads and then from there your co founder came and worked with you and you guys said, this is it, we're out of here.
Abdulrahman Jamjoom:
Yeah. So Rami and I always knew that we were going to start a company together. So while I was actually at Threads, we started, we were trying to build a browser and that didn't end up working, but like, we really enjoyed working together. So he ended up joining Threads. We're working at the same team and really early on, once he joined, we kind of knew that we wanted to do something, we wanted to do a startup together. So we started researching a lot of different verticals, like different markets, just to get a Sense of, like, where are the real problems in the world? obviously we wanted to fix a real problem, like, a real pain that people are facing. So we talked to charter bus companies, lawyers. We're thinking of certs, like, so many different.
Abdulrahman Jamjoom:
And it took us a while, like, months. We're talking, just like, researching. When we got into healthcare, we were fascinated by how chaotic the front desk is. Like, truly, you enter and it's so anxiety inducing. You're getting calls, messages, voicemails, fax. Fax is still being used. People waiting on the line to, like, check in. It was insane.
Abdulrahman Jamjoom:
There's so much. I remember one. One time I went to this general clinic in San Francisco the day I was shadowing them just to learn a bit. Two of the front desk quit that day. So there were total four. Two quit. So I actually had to sit and assist them in figuring out how to triage all these notifications that are coming in.
Demetrios:
Oh, man.
Abdulrahman Jamjoom:
So, yeah, it was super chaotic. And that's really where we started thinking of Arini. What can we do to help solve this problem? That's the initial phases of the idea.
Demetrios:
You just glossed over. Something there that I would love to get a little more context on is how did somebody let you into their clinic and just have you hanging out, working with them on the first day when you're shadowing?
Abdulrahman Jamjoom:
So, yeah, I mean, you kind of just walk and knock and talk and until you find the right person. I mean, not everyone agreed to us doing it. We definitely got shooed away by a lot of practices.
Demetrios:
And what was the ask? Can we just hang out?
Abdulrahman Jamjoom:
Yeah, can we learn? Like, we're, our background is in this and that. We want to start a company. We're researching different verticals. Can you just allow us to shadow you for a few hours? And we actually found quite a few groups that, fully gave us, full access. They're you want to go to the back? You want to see how we do stuff in the back office? Yes, go. you want to stay in the front, Go. Because they had, we sat down. I remember this very, vividly.
Abdulrahman Jamjoom:
Literally. Ram and I were working on Mission Street. I'm not going to say the name of the practice, but it's like a big group. We knocked on the door. The owner was in there on Saturday. the practice was closed, and he was grinding, through paperwork. We went, we talked to him, and we're hey, we're interested in doing something for healthcare. We don't know what can you teach us? And we sat down for a few hours and he told us all his problems.
Abdulrahman Jamjoom:
He's back office problems, front office problems. And at the end we're can we just see ourselves can I come on Monday? Once you open and see in practice, he's yeah, absolutely, I don't care, like come. And we did that. And that was honestly like a game changer to us because we saw it in real life how chaotic it is. And we did that a few more times until we build up, built a hypothesis which is there's huge staffing shortage. It's really, really hard, hard to hire people on the front desk. And even if you hire, there's a lot of churn. So you hire someone, you train them, you spend so much money and time training them and then they leave like a few weeks later, a few months later.
Abdulrahman Jamjoom:
So you have to find someone else.
Demetrios:
Yeah, because it is so stressful.
Abdulrahman Jamjoom:
Yeah. So that problem very much excited us to solve it.
Demetrios:
You know, it feels like there is a need there. Much more. It's very much like you're selling painkillers versus vitamins as opposed to the browser, I think.
Abdulrahman Jamjoom:
Yeah. So interesting thing. So after working on our first startup, when we were thinking of like the next startup we wanted to do, we put a few criterias that were absolutely crucial for us to hit. First thing, it has to be actually a painkiller, like a real problem. If Arini doesn't exist, people are suffering sort of thing. You wanted to have very clear monetization strategy and very clear go to market strategy. And with our previous startup we had none of these. So we really focus on figuring out go to market monetization and solving a real problem, like a painful problem.
Abdulrahman Jamjoom:
And that's where Irini came. The market is ready for it, the technology is there. And sure, it takes a lot of effort to make the system that we're building reliable. It's scalable. But we've built reliable and scalable systems before and that's what we think we're really good at. So we took that problem and started cranking at it until we started getting clients and expanding.
Demetrios:
Yeah, so you end up saying there's a lot of pain here, we need to build something. It's the front office, there's a staffing shortage, it's anxiety inducing. How do you land on Arini and what it does today and what exactly does it do?
Abdulrahman Jamjoom:
So some context there. So Arini is the complete AI solution for patient communication. Our focus is serving Dental practices primarily in the US and Canada. We mainly work with DSOs. These are dental support organizations, basically a group that owns many locations, many dental practices. So we're talking anything between like 10 to 100. So the solution that we offer is an agent that analyzes phone calls and text messages. So it is omnichannel.
Abdulrahman Jamjoom:
We're not just focused on the voice. We're meeting patients, where patients are. And the goal is to schedule, reschedule, cancel, confirm, like really replace that whole communication cycle that currently either a call center does or the front desk has to do. And they spend hours doing it every single day. So that's. that's what the product does.
Demetrios:
Why did you go after dentists and not the doctors?
Abdulrahman Jamjoom:
So that's a great question. So the dental market is really interesting in the sense that there's a lot of consolidation happening right now. So these DSOs that I mentioned earlier, they're acquiring a lot of smaller practices and putting them under one umbrella. And this is quite unique to the dental industry. There's also like similarity in optometry, for example, and a few other healthcare verticals. But dental has a very high growth rate for these mid market DSOs, like the 10 to 50 locations that want to reach 100 locations. So for them to scale to that extent, they have to hire so many people. And with the staffing shortage, that's pretty much impossible.
Abdulrahman Jamjoom:
So Arini solves that problem by offering them a solution that handles all of communication for them. You don't need to hire four people in the front desk, you need to hire one person to greet real patients that are coming to the office. And then we can handle all the communication that doesn't happen face to face. And we've seen that in practice. Like there are many groups that are using Arini not just as a call overflow system. We're answering 100% of their calls, like truly 100% of the calls, booking 700 to 1,000 appointments per month per location. And that's like one location from the 20 that they have. So like the ROI is crazy that's why a lot of people are starting to get into this like industry.
Abdulrahman Jamjoom:
Competition is rising. That's nothing to hide. There's it's such a clear problem to solve. You just have to solve it reliably, effectively. You need to scale. Well, it's a very eng heavy problem to solve.
Demetrios:
Yeah. Talk to me a little bit more about some of the challenges that you feel you hit as you are building out these voice agents that can scale and can be useful to both the dental companies and the end user who's having to talk with them.
Abdulrahman Jamjoom:
Well, the thing is the quality of voice in general. Like, we're talking transcription, LLM synthesis. If you're thinking of that, whole cycle, now we have real time API. It improves things. Or like, that's the next level. But like, when Arini started, we were building that pipeline and we use that pipeline until now, actually, the complexity there is okay, how do you do interruptions properly? How do you reduce latency? How do you make sure that when someone says their insurance, you actually write it down properly? Because if you don't write it down properly and you book an appointment, the front desk is going to have to call the patient and get the information you get. So a big part of Arini doing things reliably, that's what matters. Building a demo is not too difficult, actually making it scalable and reliable to take thousands of calls every day.
Abdulrahman Jamjoom:
That's where things get kind of crazy. So we spent a lot of time on the infrastructure making sure that we can do that before scaling. I think humans in general have been trained to absolutely hate talking to bots over the phone, including me. I'm not going to lie, everyone for the past 10 years, when you call your bank or the airlines, it's so horrible, so painful. if you hear not a human on the other end of the line, you'll just hang up, you know?
Demetrios:
Yeah, we yell human. That's my. That's my go to. I just start yelling human, human assistant or operator.
Abdulrahman Jamjoom:
You know, just, yeah, operator, leave me alone. I don't want to talk to an AI, you know? Yeah, so how do we. Sure, the technology has improved, but people still have that perception of AI is not there. Like, these bots are stupid and slow, blah, blah. So we actually spent a lot of time on trying to figure out how do we give the end user, the patient who's calling, enough confidence so they at least try it out, so it comes down to that initial greeting. Hi, thank you for calling Arini Dental. How can I help you? Is it better to say how can I help you? Or who do I have the pleasure of speaking with? Or should we add background noise? Or there are so many different variants. Because really what we want is we just do not want them to hang up when they hear the greeting.
Abdulrahman Jamjoom:
We want to convince them that this AI is smart enough to handle their requests. So actually we recently ran AB test and we found that it's a lot better to ask for the caller's name in the beginning. So, hi, thank you for calling. Smiles Dental, who do I have the pleasure of speaking with is a lot better than, hey, thanks for calling Arini Dental. How can I help you? The reason being you're giving them a very specific question. They don't have to think what is your name? Oh, my name is Jason. Oh, hi Jason, how can I help you today? And that already showed them that the latency is low. It repeated your name properly.
Abdulrahman Jamjoom:
It sounded good. Let me try it out. I think that's the hypothesis that I'm building around this that's why this greeting performed better.
Demetrios:
Sidebar. Smiles Dental, if that doesn't already exist, great name for one of these 100 person or 100 location dental spots. The thing there's a lot of pieces here that I think about all the time, specifically when dealing with agents. And are you calling out that it is robotic or it is AI at any point, or do you just let the end user figure that out?
Abdulrahman Jamjoom:
So we make it easy for our groups to just decide if they want to add that like announcement, hey, this is, whatever name the practice is named, this AI agent helping to book appointments, so on. So we can do that. We by default don't start doing that because we've actually seen it doesn't really change the conversion much like the success rate. And I think it just delays getting to the point. People are calling to just book an appointment and if you get them that appointment in under like two minutes, they don't care. They genuinely do not care if it sounds robotic as long as it does the job, And I think that actually something probably that differentiates us from a lot of people, the optimization is not just make it sound as natural as possible, it's get it to do the job as fast as possible. no one wants to be on the phone for seven minutes. Even if it sounds perfectly like a human get the job done, like including me.
Abdulrahman Jamjoom:
I don't want to talk to an AI for like seven minutes. I just want my appointment, give it to me. And I think that like really helps us figure out what work to prioritize.
Demetrios:
You know, when you're plugging into these dental groups and you recognize that Irina does that front end or that customer relation, are you also getting access to plug into the systems of the dental groups?
Abdulrahman Jamjoom:
Yeah, absolutely. So part of it is we have to integrate. So it's called the practice management system. So that's where they schedule appointments. That's like thinking it has the CRM and scheduling of or dental practice. They have the X rays and all that information. So we have to connect to these systems. And actually that's something that we spent a ton of time on in terms of being able not just to connect to practice management systems, rather be able to connect in a way that allows us to schedule very complex appointments.
Abdulrahman Jamjoom:
So different groups, for example, have templates on their schedule. Some days they want you to stagger appointments, some days they have extra assistance so they can double book appointments. So that's when things get very complicated. Because when we're onboarding a practice or a group, they're basically onboarding us as if we're a new like front desk receptionist and they're teaching. This is how emergencies are booked, this is how cleanings are booked on these days. You need to stagger on these days, use these templates. So we have to take into account all these settings and we invested a lot of like infrastructure work to be, to be able to support that at scale, so you can go and edit your settings and it instantly knows how to schedule appointments properly for you, because that's a big part of it. If we're putting appointments in the wrong spot, then we're not doing anything useful.
Demetrios:
Yeah, you're creating more work because then they got to go back and reschedule or they. Exactly. And the end user is going to be frustrated because they came to what they thought was an appointment only to find out that they're double booked and then they have to wait longer or something like that.
Abdulrahman Jamjoom:
And you know, something interesting here is that different groups can use the same exact scheduling software completely differently. For example, templates for one group mean that you should book here, for another group means that you cannot book here, for a third group means that some, depending on the title of the template, you can book here, some you can. So it kind of has to be a very flexible system that you configure and doing that at scale was quite complex. But I think we did a really good job with it. So now like onboarding is like very streamlined. You get set up, get started, test out the product and then you go live taking real patient call. It's not actually just calls, it's calls and complete text message support. So if someone texts, they want to cancel, confirm, they can do that as well.
Demetrios:
So at Smiles Inc. You've got root canal Wednesdays and those only days that you get to schedule root canals just to make see how versatile this agent really is.
Abdulrahman Jamjoom:
Yeah, actually yesterday during Christmas. I like this feature. It's called the knowledge base. So we used to internally maintain each group's knowledge base, but then yesterday we decided to open it up to everyone to tune it, but with very good educational content. How do you do prompting properly for a group? You want to make sure you prompt that, you test it. So like walking the user throughout that whole steps of okay, you made this change, great, now test it before you deploy it. So very excited about that. I think that type of work will really help us scale because rather than going through support and talking to a human, you can just like do it your own.
Abdulrahman Jamjoom:
Learn how to prompt your agents your own and like fix any issues that you hear from the calls that you're getting.
Demetrios:
You're teaching dentists best software engineering practices. I love this.
Abdulrahman Jamjoom:
That's future.
Demetrios:
I've got one more question before we go, which is around why you think in the US specifically today, even though it is possible most of the time to book appointments online, people still want to call.
Abdulrahman Jamjoom:
Yep. Because it's not just about booking the appointment. There's a lot of questions that go around it. You're asking about insurance, how long is it going to take? Do you do the service? Do you have parking? We actually get a lot of questions about parking. Do you guys have parking in the back? can I leave my clarification questions. So rather than just like booking the appointment, you're asking questions about it. You're asking things maybe online you couldn't find a slot in the morning, you want to inquire is the doctor in in the morning? Can I just like go as a wait list or things of that sort? And with an agent, you can train it to do these things. You have the knowledge base that you can update and say, if it's an absolute emergency, just show up.
Abdulrahman Jamjoom:
We'll assist you even if we can't book you in. So things like that people call for. We get a ton of calls in the morning, during lunchtime and after hours for the practices that are using Arini as an overflow. And then of course there are the practices that are using it for 100% of calls. But these are like the peak hours that we see. It's quite interesting to see the graph.
