2026.09.25 | GP | MedTech | Rad AI Customer Review

Workflow Platform

Primary:

RAD AI, Inc.

Associated:

Sirona Medical, Inc.

Aidoc Medical Ltd

DeepHealth, Inc.

Gleamer SAS

Circle Cardiovascular Imaging Inc.

Thirona Bio, Inc.

New Lantern, Inc.

Radiomics

Viewpoint:

Customer

Moderator: Vithika Agrawal (VA)

| Sep 24, 2026 | 21 Min Read

Amit Mehta (AM), Chief Executive Officer - Amplifier Health

SU MMARY

The discussion focuses on Rad AI, an AI-powered radiology software platform designed to enhance radiologists' workflows. The expert, a leader in health technology, shares insights on how Rad AI integrates into their radiology practice, which handles a substantial volume of imaging studies daily. The conversation explores the selection process for adopting Rad AI, the challenges faced in radiology that necessitate such tools, and the platform's value proposition in terms of efficiency and cost-effectiveness.

Key Insights

The expert emphasizes the significant increase in imaging volume, which has driven the need for AI tools like Rad AI to improve efficiency and manage workload. Rad AI is highlighted for its ability to save radiologist time, thus reducing the need for additional hires and providing a direct return on investment (ROI). The expert notes that while the technology itself is not unique, Rad AI's integration and compliance features make it a practical choice for radiology practices, especially smaller ones without the resources to develop their own tools.

Competitive Landscape

Company/Product

Positive Aspects

Negative Aspects

Market Position

Rad AI

RadPair

Increases radiologist efficiency, HIPAA compliant, integrates well with existing systems

Technology is easily replicable, no unique technical moat

Strong in reporting and impressions, suitable for practices of varying sizes

Similar offerings in voice reporting

Newer to the market, less established

Comparable alternative, potential future competitor

Other Platforms

Offer comprehensive reporting solutions

Higher cost for full solutions

Emerging competitors, but not yet fully evaluated

Quantitative Insights

Metric

Value/Insight

Efficiency Gain

20-30% time savings per radiologist

Pricing Model

Per radiologist seat or per study

Adoption Rate

Increasing industry-wide

The discussion provides a comprehensive view of how AI tools like Rad AI are transforming radiology practices by improving efficiency and managing increasing workloads. The expert's insights suggest a growing trend towards adopting such technologies, although cost and ROI remain critical factors in decision-making.

VA: An analyst here at Guidepoint Insights, welcome to our Rad AI call, which is an AI-powered radiology software platform that helps radiologists automate and improve their workflows. Our advisor for today's call is Amit Mehta, who is the Chief Executive Officer at Amplifier Health, where he leads a company that designs and develops a platform that detects human health conditions with sound. Amit, thank you so much for joining us today. It's a pleasure to have you on our call. Could you please start off by providing us with a summary of your professional background?

AM: Sure. I'm a radiologist, work in a practice that is a large local regional footprint. We have 100 radiologists, 36 imaging centers and 15 hospitals doing about 8,000 to 9,000 studies a day. We have a full tech stack, about 17 products across efficiency, AI, transcription and all of radiology. We use 7 different PACS systems across different vendors because of sort of disparate owners. It's an inpatient and outpatient practice, a local practice, a local regional practice and a remote practice. So there's teleradiology, on-site radiology across both radiology-driven as well as non-radiologist-driven workflow.

VA: Thank you for the background, Amit. Also, just to kick off our discussion, could you maybe talk about the process that your organization took to select a platform like Rad AI?

AM: So we do an RFP for different functions once a year. That RFP is solicited to all the vendors that are in that space based on whatever it is, AI, transcription, PACS. We have a specific set of metrics that we generate that are looking at things like turnaround time and radiologist time efficiency that we've developed over the years for each of those sub- disciplines into whatever tech. We put that RFP out for process. There's 60 days to respond. And then we have a selection process that then goes 90 days. We have a committee of 7 that has representation from inpatient, outpatient, hospital, CIO, CMO, CFO. I run that committee. I'm the sole decision maker at the end. But I do solicit input from the other 6 people. That process runs 90 days. There's a recommendation made to the practice to the change, depending on who the stakeholder is, and then we typically allow 30 days to implement a change.

VA: Got it. And did you previously use a platform like Rad AI or was this like one of the first platforms that your organization adopted?

AM: No, it's one of many. I mean, we have 17, so it's one of the 17 that we use.

VA: So can you highlight some of the challenges or maybe pain points that has led you to kind of seek a solution like this?

AM: So in imaging, specifically in radiology, there's an increase in volume, which is the biggest driver in that the way

the healthcare system is set up is that it's become a lot easier to order imaging studies and they're a lot easier to get done. So our volume has increased above baseline. In terms of just population growth and almost explosive, we can't keep up as the radiologists with the volume, hence lots of AI tools being developed. We look for tools. There are 2 issues with these tools. One is we're looking for obviously a clinical benefit. Does it make us better as radiologists? Does it find a fracture on an x-ray? Does it detect a tumor? But then there has to be an ROI, right? Because there's really in the chain of payers, people who pay for these things, there are 3 big buckets of places where you can get money to do these things. There's the payer, the provider, or the patient. The payers aren't paying for anything, there's no CPT code for CAD except for mammography, right? So it's almost impossible to get a payer to pay for any of these technologies. The patients don't want to pay for anything. I mean, we've implemented 3D breast and patients won't pay for it; if they have to pay out of pocket, they opt out. And so the last person left is the provider. We get paid $7 to do a chest x-ray, we're not going to pay $5 to a CAD system that's going to detect a nodule, right? So there really is no ROI on most of these technologies. And that's the... the block for Rad AI. The way they find is that they're actually building tools for the radiologist to make them more efficient. And so the way it works is by using Rad AI, we can be more efficient in what we do and decrease the number of radiologists we need. And that has a direct ROI. So if we save 30% time on the amount of work or the amount of effort it takes to do 100% of the work, then, you know, obviously every 3 radiologists we can save hiring a fourth. And that's how we save money using the tool.

VA: And can you maybe talk about the core value proposition that Rad AI has provided for your organization?

AM: Can I talk about what?

VA: The value proposition that Rad AI has provided for your organization.

AM: So they have a set of reporting and follow-up tools that provide increased efficiency for the radiologist. And that increase in efficiency for the radiologist allows us to do more work per person. The radiologist is the largest expense in the chain of imaging for us as the provider, not in the system. Remember, the system, the largest charges, the technical fee that the hospital charges to do the study. But for us as radiologists, we have a bunch of tech fees, a professional fee. As a practice, our business expenses are humans. This allows us to have less people to do a larger amount of work. Now, I mean, the underlying is that we're still hiring as many people as we can because the volume is so great. And you make it up, whatever the number of efficiency is that Rad AI will tell you. But let's say you get 10 or 15 or 20% efficiency, even at that increased efficiency we still don't have enough people. But the value proposition is increased efficiency per radiologist to get more studies done.

VA: Very clear. And are there specific product offerings that your organization has purchased to achieve this?

AM: What was that? Sorry, it's cutting out a little bit for me.

VA: Sorry, is it clearer now?

AM: Yeah, it's in and out a little bit. That's fine, I can hear you now.

VA: Okay, I was saying, are there any specific product offerings that your organization has purchased to kind of achieve that high level of efficiency?

AM: Yeah, so in this tech stack of these products there's a whole set of tools, right? So there's a set of tools that make imaging more efficient. There's a set of tools that make the reporting more efficient. There's a set of tools that make the follow-up more efficient. There's a set of tools that make the communication more efficient. So across the processes that we do as radiologists, across the interpretation and management of these exams, there's different AI tools that do all of these things. Rad AI fits very specifically in the transcription reporting part of this.

VA: Got it. And when it comes to Rad AI, which were the other vendors for the workflow specific to Rad AI, which were the other vendors that you ended up evaluating during the sales process?

AM: The other, what did you say, the other vendors?

VA: The other vendors, the other platforms that you ended up evaluating. Yeah.

AM: Oh, there isn't, I wouldn't say there's a specific competitor. I mean, there are, there's RadPair, there's a few others, but we didn't evaluate, we were using Rad AI. We were using 1 of the Rad AI, the impression product already. So this was an extension of the current vendor for the most part to add in Omni and add in these other products that they have. So we did a very cursory, but the other sort of generative AI LLM products out there were relatively new and we sort of just decided we'll wait 1 cycle before looking at them deeply. They're just sort of coming to market. The true competitors like RadPair, the ones that exist as a competitor, we're just waiting for another cycle.

VA: Now, as you said, there are multiple platforms out there that do similar things and Rad AI just focuses on 1 particular workflow. What are the features of Rad AI that you consider are its biggest differentiators?

AM: So there's 3 products. There's Rad AI reporting, Rad AI impressions, and Rad AI continuity. Start with the continuity. The continuity product, I don't think it's a differentiator. We don't really use it or pay for it. It provides the ability that when there's a finding on a study, it will do communication with the provider, the nurse, whomever, that this patient had a finding, you need a follow-up X, Y, Z. We found that that doesn't really happen in our environment. That doesn't really... either the doctor provider does follow up and they don't need a reminder from us, or that patient is lost to follow-up. So for us, we don't have an ROI on that. Like there is a sort of high-level ROI that whatever it costs to run Rad AI continuity, if those patients do come back for a follow-up study because you reminded the doctor that the extra billing would be for that study, but we're having trouble keeping up with the studies that we have, so we're not looking for superfluous volume in that kind of way, so we don't really use that product... the other two. Impressions is an efficiency gain. The most taxing portion of a radiologist's job when they're dictating a report is they report the findings in the study and then an impression. It automatically generates that impression and that saves us somewhere in the 20% time. And we kind of looked at it internally and agreed that we save about 20% of our time given that Rad AI impressions can generate for us. So that is our workhorse product. And then there's a Rad AI reporting tool, the Omni tool, that is a sort of generative AI, check or follow-up, builds the reports a little. It's impressions for the entire report. That's similar and does similar savings in time depending on what you're reading and how detailed you are in that. Like for a chest x-ray, for example, it's not worth paying money for an AI tool that does this kind of stuff because most of them are normal and the reports are very brief. This is for higher value cross- sectional imaging like MRI or CT, where there's a longer report and multiple findings and multiple follow-ups and all this kind of stuff.

VA: But is there a risk that... the competing platforms... the other platforms that you're using adjacent to Rad AI, could they also expand their offerings to kind of replicate Rad AI's workflow capabilities?

AM: I don't think there's anything. There's no moat around what Rad AI is doing. Anybody could, I mean, honestly, you could "vibe code" what they do now in 5 minutes. It's not so much the tech anymore. When they first came out with it, yes, but now we've built some of these tools internally ourselves, like "vibe coded" them ourselves. So the technology is not so much; it's sort of their distribution is more what's novel.

VA: So there's no differentiation as such. It's largely commoditized and quite similar is what you're saying.

AM: Yeah, I mean, honestly, if you wanted to try it, you could go into ChatGPT and say go to Rad AI and make this impression thing for me, and it'll do it. I mean, you can vibe code this. So it's

VA: Go AM: Not... there's no difference. I don't think there's a differentiation in the product itself, but they have a platform that integrates into the reporting system and they built the connectors and they have the contracts and the business agreements and all that kind of stuff. I mean, that's where the business is, right? It's HIPAA compliant, SOC 2 compliant, and all of that kind of stuff. So if you're a radiology organization, I mean, we're a big group, right? We have 100 radiologists and we have a full tech team because we support ourselves. We have the wherewithal to create our own tools and do this, but like, if you're a 3, 4, 5 radiologists practice in Tulsa, Oklahoma, it's unlikely you're going to try to build these things yourself. So you buy them from Rad AI.

VA: And what has been the key purchasing criteria for your organization to choose Rad AI as a platform?

AM: That ROI and savings on the radiologist time.

VA: Got it. And before we talk about the ROI, can you maybe tell us about the pricing, just on the pricing front, how do they price their platform? What is the pricing model that they use, etc?

AM: Yeah, I can't tell you specifically what our pricing is because we have an NDA because we were a development site for them. But there are 2 sort of methodologies here in general. And this is true in the PACS environment as well, is that there's a per study fee. So SaaS-like, but it's per study. And so if you're a low volume place, there's no upfront cost, you use the SaaS fee. And then other than that, it's a license per radiologist fee. So we're on the license per radiologist because we do such a high volume. And so per radiologist, we pay for a seat.

VA: And is there any difference between how Rad AI prices their product versus other similar platforms?

AM: No, it's essentially most people are doing it per seat. For a long time, people tried to do per volume, like in PACS world, they did that too. It was a per click study, and so they would say, "Look, this is the proposed growth over the next 5 years," and then there was a fee if you went above that. It turned out we blew through all those numbers, right? What people thought was going to be sort of 3.5% to 4% growth turned out to be 15% growth. And so the provider side very quickly went back to vendors and said, "Look, we're not doing this per click thing anymore. Either you sell us a one-time solution or we'll go to someone else." And there's always a provider who'll do it as a per license, per seat, per practice, per kind of thing. So people moved to that model very quickly.

VA: And when it comes to like deploying the product, what was the level of effort and time that was required to deploy Rad AI's product onto your systems?

AM: Pretty low lift in the world of radiology today, all of the systems are very modular. Everything is built from the ground up in order to be able to use these tools. So we have AI doc, we have Rad AI, we have a bunch of tools, primordial, that all overlay on the PACS and the reporting system, but they're all very modular and very quick.

VA: And now like we've spoken about efficiency a lot. So can you maybe talk about this in terms of any ballpark numbers? I think you mentioned it initially, but what has been like the ROI that you've seen so far from working with Rad AI?

AM: It's typically, per radiologist, purported between 20 and 30%.

VA: And how does this compare with other competing solutions? Is it on par or are there solutions which offer better ROI?

AM: They're all about, yeah, I think they're on par. So in terms of a reporting structure, so the difficulty with that question is they're all a little different, right? Like, so Rad AI has Rad AI Reporting and Rad AI Impressions. We use

Impressions more than we use reporting. RadPair and some of these other ones, what they provide is the full reporting solution because there's an LLM underneath. So it's a different, they're different technologies, but they're similar. Like, you have to get fairly detailed into the weeds about what the difference between these things are, but for us, for this Impressions product, which is what we pay for and use for the most part, versus the Rad AI Reporting and RadPair, it's hard to compare them because what Rad AI Impressions does, the other tools do all in. They do that as part of their offering as well, but then they charge more. So that's the difference.

VA: Can you talk about the scalability aspect? How scalable is this return? Has it increased or has the ROI increased over time or has it remained stable?

AM: I would say it stayed stable in that. I mean, what it does and what it's able to do is the same on 1, 2, 10, 100 studies. It just depends on how many you are reading that day. So it gives you that same level of efficiency whether you read 10, 100, or 1,000. So that is not changed.

VA: And can you talk about the level of customer service that Rad AI provides and how has that changed since the initial onboarding?

AM: These tools, once they're deployed, are pretty self-sufficient. We don't need a lot of support, so I don't think we've ever called them for support, to be honest. I can't remember the last time. We have our own tech team; they're facile in deploying it, using it, all of that kind of stuff. So our tech people are the ones who help us with what we do. We don't need to talk to these folks pretty much.

VA: And do you believe Rad AI can scale alongside your organization as your organization grows in size or will you need another vendor in the future?

AM: No, for what it does, I don't think it's an issue.

VA: And I think you mentioned RadPair, but if you compare all the alternatives, which vendor do you think is the strongest alternative to Rad AI?

AM: I mean, if you look at sort of what they all do, RadPair is probably the most analogous in that it is a voice reporting system. And so from the perspective of a competitive product, that's probably the closest.

VA: And are there any newer platforms also that are coming in this space that might kind of start competing with Rad AI?

AM: Not that I know. There are several automated radiology reports that all will provide a report when you use their MRI, PET, CT product. Sirona does reporting internally, so there are, but it's not like Sirona; you're buying a PACS, right? So you get a PACS plus the... it's not that it's specifically to... just the reporting portion. So in this reporting portion, there are probably RadPair, RadViz, and Rad AI... probably the 3 sort of bigger ones.

VA: And on a scale of 1 to 5, with 5 being very difficult, how challenging is it to switch between platforms? Very easy. So do you anticipate to move—as you said, RadPair is quite strong compared to Rad AI. So what might prompt a switch to RadPair in the future if you might switch?

AM: I mean, it would come down to cost. If we get the same features and it's cheaper, then we would move. And right now the way we're set up it's not... it hasn't played... I mean, like I said, for the Rad AI product, we haven't double clicked that much into RadPair or Fluency or RadWays or any of these other ones yet. We will in the next cycle as these ones kind of have probably... I think RadPair doesn't even have a year under its belt as a company so.

VA: So, as you said, given what you know right now, you will continue to use a Rad AI solution, even when there is an opportunity to renew given the current pricing, right?

AM: Yes. Well, our contract does not renew. We would go through the next RFP process before the renewal, so we will check before we renew.

VA: So, right.

AM: No, it won't be a default that just because we're going to just continue with Rad AI. We will do an evaluation before we do anything.

VA: And in general, how price sensitive are organizations towards platforms like these?

AM: For this particular situation, we're very price sensitive because of the fact that we're paying for it, right? This is what I was telling you, there's providers, payers, and patients. We as a provider have to pay for this ourselves. So we are very sensitive to cost.

VA: And have they revised pricing in the future or quite often or has pricing been stable?

AM: Stable.

VA: Sorry, did you say they will?

AM: It's stable right now.

VA: It's stable. And what has been the adoption rate? If you consider the broader industry, what is the adoption rate for such a software provider?

AM: I think it's probably pretty different for different organizations in general. For an organization like ours, I told you the process, we do this RFP, we do evaluation, et cetera, et cetera. I think for a smaller practice, it's probably a lot easier, the sales cycle. They just started, I think the salesperson can present, they can show an imputed value and then given the contract size and the ease of turning it on and off, they can just try it. And if they don't like it after a certain period of time, just turn it off. It's sort of a subscription kind of base thing, right?

VA: And how has this adoption essentially evolved over time?

AM: I mean, for us, we adopted it and now we use it. So I don't, is that what you mean or what do you mean by that? It hasn't changed, right? We adopted it.

VA: The broader industry, has the adoption rate increased towards platforms like these?

AM: Yeah, I think they have, because if you look at it, I mean, these guys raised a $68 million round in January of 2025, right? So someone, I don't know their underlying numbers, but presumably someone's funding that kind of funding because people are paying for it.

VA: And when it comes to these platforms like Rad AI, RadPair and some of the other platforms as well, do you think, can you rate them on a scale of 1 to 10 with 10 being the highest?

AM: For what criteria?

VA: Based on the overall platform performance?

AM: I mean, they all sort of are 10 out of 10 for what they do, right? It's a very specific sliver of what it's doing. So if you said, how good is Rad AI at transcription? Well, it's not really a transcription product per se. So it's not like Rad AI as a product doesn't really do that. So for their specific ones, they're all kind of do what they're supposed to.

VA: And is there any features or improvements that you'd like to see in future versions of the product?

AM: No, I think it's a good question. I mean, they're developing out a bunch of stuff. I think the horizon of products are good. It just has to be very focused on ROI for the radiology group, like the discontinuity product, for example. We can't find value in that in terms of paying for that. So in that case, it's not something that we would do. But if they can find ROI value, then they're great products that we would use.

VA: All right. So Amit, I think we covered a good amount of ground. Is there anything important that we didn't touch upon that you'd want to highlight about Rad AI or the space in general?

AM: I think from a commodity standpoint, these tools are becoming easier and easier to create, and I don't think there's a technical moat around what they do. Like I said, in our practice, a couple of these tools, we coded ourselves and we use our own tools internally. We don't have to, we don't need sort of FDA approval, so to speak, regulatory approval if we're just using it for our own purposes. And I think that is going to become more prevalent than buying these solutions. And again, this is for large practices that have the capacity to do that. But I think that's going to become more commonplace.

VA: Great. Thank you so much, Amit. This was very informative. Thank you for your time.

AM: Okay, awesome.

VA: Bye-bye. Have a great day.

AM: Bye. See you.

DISCLAIMER: The statements or opinions expressed today are those of the Advisor and not Guidepoint, who disclaims all liability for the content provided. The Advisor may not disclose material nonpublic or confidential information or any information that would cause the Advisor to breach any duty or obligations. Guidepoint is not a registered investment advisor and the information provided is not intended to constitute investment advice.