Steve and Sam are joined by One Chowdhury, founder of the AI-native CRM, Octolane. One shares his incredible journey from a Duke University student who didn’t know what a CRM was to a Y Combinator-backed founder aiming to take down Salesforce.
Listen to this episode here, on Apple Podcasts, Spotify, or wherever you get your podcasts.
He discusses why legacy CRMs are “trapped with their own success” and struggle to adapt to AI, while Octolane is built from the ground up to eliminate manual data entry and give sales reps their time back. Learn how Octolane’s ‘Action Mode’ proactively suggests follow-ups and meeting prep, aiming to turn every user into a ‘superhuman seller’. This conversation covers the future of CRM, go-to-market strategy, and building a product that solves the core pains of sales professionals.
Transcript
STEVE: Welcome to CRM Talk, the show that brings you the latest in CRM and CRM related news and information. This is Steve Chipman along with my co-host Sam Biardo and today we have a special guest and that is One Chowdhury, founder of Octolane, an AI native CRM. Welcome One.
ONE: Thank you Steve. And great meeting you guys. Really, really appreciate for giving me the opportunity to be here.
STEVE: Glad you made it. So my first question, my burning question is, what drove you to develop a CRM platform with all the tech that’s out there?
ONE: So, frankly speaking, I actually didn’t know what a CRM was. I born literally yesterday. It’s very rare that a college kid would build a B2B SaaS out of like everything. So I’m actually class of 25 from Duke University. I dropped out during my junior semester to build an AI CRM to take down Salesforce. And I also didn’t know what Salesforce was. So basically, I did my summer internship during sophomore year in San Francisco at a, at a YC backed startup named Mintlify. Now they are like a16z backed. They’re at the AI documentation. So I was just working there seven days a week. I would not go out. Like everybody was working hard. It was very early days, like co-working space. One day for some reason I did go out in SF, I saw the tallest tower in SF and there was like a guy, my friend with me, he was doing internship in LinkedIn. I asked him, what’s that tower? He’s like, that’s Salesforce. Everybody hates them. And I am like, if everybody hates them, why they have the tallest tower in SF? Then I am like, okay, that’s the tower that was built on pain. Then I started learning more that people actually hate Salesforce, but they tolerate HubSpot. There’s this weird dynamics. People hate it, but they still use it. And it was like a moment, you know, like in 2023. So AI just came into the scene. I was also building like an AI features for Mintlify. And I’m like, why people cannot build something like better with LLM? Like manual data entry is just a huge problem. And I then I was like, if nobody else is building a new CRM, maybe this problem is solved, you know, then maybe I should not. Then I found this company named Atio, and they just did their series A and at that moment, now Atio is like an AI CRM, but at that moment they didn’t have any AI built into it. So I was like, this is the opportunity. So I started building. It was, it got like immediate traction. So I was in a co-working space in Mission Street. There are like multiple founders, like we were all building together. Founder of Exa AI, it was called Metaphor AI at that time. Founder of brave.dev, they got acquired by Nvidia later. So you can see these people and they would hate, they, some of them were using Notion as the CRM, spreadsheet, their Apple Notes. So they would use then like HubSpot, they would hate it. Then people would say go to Salesforce, you find your, you will find, you know, like it’s like good solution. They would go there. So my first version was a very basic script where it would just see the anonymous traffic of your website and use the IP address and the geolocation and try to match that with office address of like tech companies and try to identify what companies are visiting your websites. Then there is like an agent that would go to LinkedIn and find people and would just like send emails for you. Sounds like an AI SDR, but then at that moment people would ask me like more stuff. It was a side project, nothing else. Like, can it work as like a contact database? Can it do like research for me? Can it automatically find accounts that I forgot to follow up in my Gmail? So then it become a CRM. So on my last day of working at Mittofy as an intern, like I asked my CEO back then like, will I get an opportunity to work here full time? I really like this culture and everything and it was really inspiring. Like people would work really hard. I think the Hun one, the CEO of Mittofy, he would be the first person to come to office and it was a co-working space and he would be the last person to leave the office. So I would be just there with him. So he said, see, you can one, you can always come to our company to work whenever you want. The door is always open. You are the best intern we have ever had and you are the first intern we had and you are the best intern we’ll ever have. But this side project that you’re building, I think it can go really big. You should apply to YC. So he wrote the referral letter, I applied to YC, I went back to college to do linear algebra, then while I was doing homework, I got the phone call from YC, did interview and everything, got in and just left Duke. My mom is really upset still. So no one understands in my family what a CRM is. So that’s bad. I wish I was building a rocket, I guess. But yeah, that’s the whole story. First forward today, there are 200 companies who are using us. We were focused mostly on switching customers from HubSpot and like other like companies rather than going for new companies. People said we are like crazy. We raised $2.6 million in funding. And we’re still building. And it’s a, I thought it’s going to take me two months, it’s been almost two years. Yeah. We are like five team members now, based in Mission Bay, and you know, I’m in this podcast right now with you guys.
STEVE: Well, and it sounds like things have come full circle. So Siebel had a system that nobody liked, but it was the 800-pound gorilla. And then Salesforce, the scrappy little company came along, and it almost sounds like they’ve gone through their entire cycle. And now you’re where Salesforce was in 1999. Is that the fair statement?
ONE: That’s, that’s exactly correct. Like nobody can deny that when Salesforce came out with their cloud architecture and everything, they were the king. Like Marc Benioff was a literally the king and like it’s a, it was a $300 billion company then what it became. Now it’s literally the full circle where a new revolution is coming, which is like LLM. Frankly speaking, if it wasn’t for AI, I don’t think any company would have a shot at like going for like to take down Salesforce. Like I talked to a lot of like board members, CROs, they are like now board are telling them to embrace new technology, like what’s AI and everything. That wasn’t the case. I think that’s like a huge unlock for us. And it’s really hard, like Salesforce is trapped with its own success, frankly speaking. Like the structured data, unstructured data, combine them, get the value out of it and just update it real time, it’s really hard.
STEVE: So One, it sounds like the challenge that client server on-prem applications had converting to cloud apps, which happened across a lot of different enterprise apps is analogous to the situation that today’s CRM vendors find themselves in trying to convert to a AI native CRM. It’s just not easy or even possible. Is that correct?
ONE: That’s absolutely correct. I think these big companies like these traditional companies like Salesforce or HubSpot, they are trapped with their own success. It’s really hard to rebuild the architecture and move fast at the same time to get the meaning of structured data, unstructured data. Frankly, I think it’s just impossible when every week there is just new model is coming up, new tech is coming up, like your entire quarterly plan is just like out of the window, not possible. So it’s only a startup by building from scratch, you can build a platform where it, if a newer model comes up, your platform will actually get better. So it’s just like, we have been seeing this frankly, like Salesforce has been trying with like Agent Force and and they are like, we have our own model and what not. So literally like two days ago, I think in Dreamforce, they are like they introduced OpenAI partnership and everything because and then we saw like this top analyst company saying it doesn’t look good for Salesforce and the stock still dropped after like Dreamforce. So you are absolutely correct on it. It’s the same cycle over again.
SAM: Well, I also think Salesforce is a piecemeal of products. I mean, if you you look at how they’ve grown, they they bought their marketing solution, they bought their BI solution, they bought their AI solution. You know, and it’s hard to it’s really very difficult to put those things all together, you know, and make them work as one. And then you lay lay your LLM on top of that, where’s it getting its data from? Which sources is it pulling it from? It’s it it adds a a new layer of complexity. So,
ONE: 100% Sam. I think it’s like a Frankenstein problem almost and they are buying startups and companies as if it’s like Pokemon. So they actually need like a Pikachu. So I think they did their sixth acquisition this year and still like going through. By Pikachu, I mean like something that moves fast, electric and like a small. But all they have like, it’s just it’s just difficult. Like what can I say? It’s like an enterprise theater right now. Like how many announcements you can do. Okay, maybe buying this would help, but it’s just not. Like this is not how it works. We all know we have seen this cycle like more than like multiple times. So yeah, you are correct on that, Sam.
STEVE: So one thing about these monstrosity CRMs, if you can call them that, they’re very embedded in the organization across different applications. You might have integrations to ERP and other in-house systems, other third-party integrations. So how do you, what do you do when companies have all of that embedded technology? How do how do you present yourself as a potential replacement for what’s in place?
ONE: 100%. It’s no secret that Salesforce or all these traditional companies actually have like two modes, right? The number one is the integration ecosystems and the second one is the consultant like ecosystem, like everybody want to use them. So the way we do it, whenever we go to company and we learn this, like we cannot tell these companies like, hey, we are going to switch you. That’s like a horrible news to them. Like they don’t want to hear that. What we say, okay, what’s the current problem that you are facing? Well, my problem is my data are not up to date. I cannot believe my own CRM. People don’t want to use the CRM because frankly it’s just so bad. The UI UX and the clicking, clicking and this dropdown hell. So what happens is that we say you don’t have to switch to us, but we have one-click login on Octolane with Salesforce and HubSpot. So you just connect to Octolane and we will keep your CRM up to date. And any, we have two-way sync between these CRMs. So anything you edit on Octolane will show up on Salesforce. Anything you edit on Salesforce will show up on Octolane. So we get like a very small department of a big company and they start using it. And after four months suddenly, okay, all of our data are already in Octolane. Why don’t we use just Octolane? So obviously we had a lot of challenges because it’s just, it’s very difficult. So frankly, at first nobody would take meeting with us, like, ah, I don’t, I don’t believe you guys, whatever, right? And then some would take meeting with us but would not believe us. Then some like started using us, then some left, then some kept staying for two months, then four months. It’s a grind. I wish there was some formula that I could have shared. There is none. We worked like more than two years and did this specific trick to with this two-way sync. So it become like a wrapper first, maintaining and mapping the objects and then go for it. And the second part is integration. We started with some specific integration like for SMBs, for meeting recorder. All we need the most important context are in emails and in the meetings. So for meetings, you will mostly use Fathom or like Fireflies. For emails, it’s mostly G Suite. So that’s how we started. And the second thing we did, a really like a big breakthrough for us was MCP, the model context protocol by Anthropic. So previously you had to build like an integration UI and all this stuff and like connect with everything. It’s just really hard. And that’s their mode, the integration ecosystem, right? With MCP, all we need just an API key. We don’t have to build an integration, an integration with Snowflake, right? We just need an API key. The LLM agent is built in a way where it can go, find the data it needs as using like its own tool and just like add the context wherever it’s necessary. So it doesn’t necessarily download the entire data from Snowflake. It just needs to like do semantic search, add the context and update that next stage or sorry, next step or update the product offering in your CRM that it needs to. That is how we’re overcoming. I’m not saying like it’s like a complete bulletproof solution, but it’s looking good for some time. Yeah.
STEVE: And do you do you ever, does Octolane ever, get adopted initially as a shadow IT application? In other words, a sales leader says, I’m just going to swipe a credit card and get my team on this and then tell everybody about it. Have you have you ever gotten in sort of that side door into certain organizations?
ONE: So when we started this company, everybody told us you should go to CRO, you should find this decision maker and try to sell to them. You know, we tried, it’s just everybody’s trying to sell to them. They’re just tired of hearing like you open and one CRO told me, I just open my phone, it’s like 20 text messages, vendors pitching me. They we had this realization after talking to like hundreds of like sales reps, CROs and like different people who are doing sales that everybody is building for managers. Salesforce was built for managers, then HubSpot was built for marketers. No one built for sales reps. So we are like, okay, we’re just going to start with sales reps. And it’s just easier. Sometimes we give the product them product to them for free, it doesn’t matter. So when sales reps uses it and when they can close faster or find some hidden deals that got lost in their Gmail, then they bring that tool to the, you know, like the upper, like upper people and then it gets adopted. So it’s very much sales rep focused. I think I, and I have like tremendous amount of empathy for this for these people. And I’m like a young guy, right? So when I started this company, it was like, okay, we are going to build the world’s largest company possible, like this SF tech, I am building an AI startup and what not. Then it became an emotional thing. As I talked to these hundreds of sales reps, there was this guy, his name is David. One day he just called me at 11:00 p.m. and and he said like he doesn’t care about Salesforce or like selling or anything. He just missed his daughter’s son’s recital third time this year because and it was like bad, right? He, he missed third year, he makes, he said he makes 400k every year, but he couldn’t attend his son’s recital because he was updating pipeline on Salesforce until 9:00 p.m. Because his CRO wants up-to-date data and Salesforce doesn’t let you update the data fast for some reason. So like the actual cost is that you are a good person, you know how to close a deal because an AI cannot replace a warm handshake and AI would not be able to replace the relationship that you’re building, the understanding. But like we have self-driving car, why we are still clicking on drop downs, right? So then I’m like, our focus should be how much time we can give humanity back. I want David to go home by 5:00 p.m. and with his son, go to that recital, go watch that soccer game, like, like it doesn’t matter. He is a good like employee, like he is trying his best, right? So then it’s like, so we have only one metric right now. Obviously we have a lot of other things is like how much time we can save for you and still make you a superhuman by doing what you do best. So that’s the only metric we track. We don’t track retention, we don’t track revenue. Internally it’s like how much time we can save. So for example.
STEVE: So you’re the ‘get your time back’ CRM.
ONE: Yes. Get your time back CRM. Like go home by 5:00 p.m. We will drive you home, the self-driving CRM. So on an average, right now, at first it was like, you know, like 20 minutes, 30 minutes per week per sales rep. Now it’s almost like four hours per sales rep. Like it’s amazing when you have that metric. Another company who did that kind of like Ramp, like how much money we can save for you and and then obviously you get time back as well because it’s just so easy to upload receipt and the AI does everything for you. So we want to give you time back and money back. The two of the most important thing.
STEVE: So I have one another question, I’ll hand it over to Sam, who I’m sure has some burning questions. So this whole concept of deep research, let’s say you’re an enterprise sales rep and you want to deep research accounts. Can you configure the system to say all my major accounts, I want you to just go out, do deep research, come back to me with all the information I need, who the key players are and present that to me as part of this account record, if you will?
ONE: 100% and it’s really good to watch because not only we give you that result, we also show you the receipts, like the evidence, why exactly we got that data. So if Octolane tells you you should call Jennifer right now, you don’t question it. You pick up your phone and you dial.
STEVE: Sam, you’re on.
SAM: Okay. Well, a first question I have is, do you build out different LLMs for different types of work? So, for example, do you have an LLM that does support cases, for example, case management and a different LLM that does opportunity management and so on?
ONE: Yes. You are exactly correct. Like it’s, it’s little difficult to use the same model for like everything. We tried that. Frankly, it would have been like cheaper and easier. It just doesn’t work that way. First of all, the moment you need different system prompts, the moment you need something like faster or something like a little different, we have to change the LLM and then you also have to use a different LLM to judge the output. So, see, we already have good LLM. So what matters right now, the context and the accuracy of the result. So, let’s say we have this web, we have a specific system prompt and specific LLM that would go and do web search for you. But to do internal search through all your records like columns and fields, we have to use a different system entirely. Then again, you have like 40 fields for a specific opportunity. I cannot like dump that entire 40 fields of data to an LLM and and ask the answer. So we have to have a different agent that will choose which input I need to take. So there’re like multiple steps. So it’s just not one LLM call. It’s like one LLM call, then another LLM is deciding, is this correct, what input I should use. Then after the output, another LLM is finding the citation for the data and like claim and the validation with evidence and then goes and it goes on and on.
SAM: So you do you have provide tools then to create those LLMs and support them? So let’s say I want to use it and I want to help my sales reps forecast their commissions. I could then take certain fields into the system, give it the formula and have it effectively build that analysis for me?
ONE: 100%. What we have seen though that people don’t want to, like they don’t care if it’s like AI or how it’s like doing the thing. They just want to type in natural language and the thing should do the thing does the thing like for them. So what we have is actually we call like description. Like you create like a column or like a field and you just type, I want product offering for this account. Enter. On the background, it will take that prompt, enhance it, try to figure out what that user actually meant, and then give you immediate test result. And if like he presses like the enter or something or edit the output, then we update the entire prompt like by ourselves. So it goes like that. And we had a section where you could have changed model, like someone wants like OpenAI, another one, maybe Anthropic or Claude. Later we realized people don’t care. So then we made it like auto mode and we don’t show them like what models we’re using like under the hood. But it’s, it’s mostly reasoning models right now.
SAM: Okay, do you then I’m assuming then you’re agentic in the effect that I can give it a command and then when I realize I’ve messed up or not provided enough information, I can go and supplement that until I get the right results.
ONE: 100%. You can edit any AI generated result and it will understand why did you or it will try to understand why did you edit that result and it will update for the next time. But if you want, you can update the entire system. Frankly, I’m not sure if there is a screen sharing here. I would just show you maybe after this podcast or during, then it’s it’s a good demo. People like it. Yeah.
SAM: Do you provide any I one of the problems I see with people building and using LLMs is they sometimes don’t have a sufficient amount of data to actually create a good model. I see this all the time where, in fact it happened to us internally. We were noticing that we would tell us that our our opportunities were all doubtful and we had no clue why because they were like in contract negotiations. Why is it telling me I’m doubtful? It turns out we were putting in some key information that we happen to be using at the time. But the one of the things I find is that a lot of companies don’t tell you how they’ve calculated their specific formulas and they don’t have any tools that will give you sort of an an accuracy, you know, number. You can do lots of analysis on the quantity of data and the quality of data and and give and calculate a probability of how good the answer is. Do you provide tools like that as well?
ONE: Yes. So, normally what we do, we give the output to users and users like tells us if he likes the result or not. And we cannot ask users like because we are not always like, you know, with them. So we just track what they do with that data. If they’re editing it, that means they didn’t like it. Previously we had a thumbs up and thumbs down, nobody clicks on those, so we just removed that thing entirely. So if they edit the answer, then we realize, okay, this is not a good answer. What can we do to make the answer better? So almost like always after one or two try, it becomes better. That being said though, when the case is super complex and the thing that you just mentioned, it does get difficult. Two data sources that we absolutely need is your like email data, so we need access to your email and the meeting data. If we miss one of those, it’s just like it doesn’t work. And LLMs are like smart enough already, the models are like good, frankly like really good. It’s just the context problem and the data quality problem. So for a lot of enterprises, it doesn’t matter how good our product is, the data they’re giving us is just bad. Like, like they don’t even know where the data is coming from, the conflict, they cannot find, there’s just so much duplicate. So when you do this two-way sync, we also have to clean the data with some like agents and like other stuff, sometimes we do it manually, otherwise it’s just, it’s just difficult and it doesn’t work sometimes. And we haven’t been able to replace like Salesforce or like big organization where they have like a lot of integrations, complex data where they have conflicting data sources even. So we started with like early stage tech startup in SF, then SMBs, and then go up and up and up. That is how currently we are handling things.
STEVE: And do you connect to Google Workspace and Office 365? Email and calendar?
ONE: Yes, so we only work with G Suite right now. Right now not Outlook. Maybe in the first quarter, probably in 2026, but not now.
STEVE: Okay, so if you’re a Google Workspace customer, this is definitely something you should be looking at. Okay.
SAM: Sam, sorry, go on. I’m I’m good, actually. I’ve run out of questions, Steve. Go on.
STEVE: I got one. So One, what was, what was one of the biggest surprises to you in terms of what users discovered through through use of Octolane?
ONE: Yes. So I think I got blinded by the fact believing that manual data entry is the ultimate like evil. Like if we can just solve that, it will, it will be the ultimate CRM revolution and everything. Turns out that wasn’t the case. Like I can tell you like your CRM is up to date, 100%, like no problem. But why does that even matter? I think it took me a long time to realize that. Sales reps, they want to close more deals, they want to send more follow ups in a timely manner. What they don’t want is like perfect data. They want like certainty. Like can your CRM help me to make me a superhuman? So it’s almost like a racing car. Like of course the architecture of that car and the engine is really impressive, but why does that matter? Because I want to finish fast, sorry, first. So the infrastructure that we have is the self-updating system, but the outcome is that you should be able to like close more deals or move faster. So that’s where we should have like focused on. So now it sounds like very basic and whatnot. Yeah, of course, like that’s the outcome. But if you don’t have that from day one while you’re building that system, you are going to optimize for the wrong things. So you are not going to optimize for the outcomes. You’re going to be super focused on the automation and like other stuff. So when we realized this and frankly we realized this like two or three months ago, it shifted the thing from like manual, we auto, we remove manual data entry to we make you a superhuman seller. So this is the only self-driving CRM that will give you superpower, which basically means, so that the messaging changes, right? Previously it was like we keep your pipeline up to date. Now it’s like we keep your pipeline up to date, but most importantly, here are four more deals that you forgot about and this is how to close them. So we built something called action mode and this is where the system of actions thing coming up, right? People are like, oh, you have to build a system of action. What does that even mean? Right? It’s a buzzword. For us, that was the unlock while we realized the system of actions mean the CRM will take action for you, but you’re still in the driving seat and you are keep approving them, editing them and you just move faster. So for us, for Octolane, the action mode is that you wake up, it will tell you five emails to follow up, pre-drafted already. You just press enter or edit, it will remember. Two meetings, meeting prep materials are ready. You do your meeting, follow up emails are ready and update CRM and everything. All you do is just press enter, enter, enter. CRM is taking action and it’s learning from you. Like after two months, it will learn your, it will know your customers better than you. That was a huge unlock. The second thing was building for sales reps. I actually underestimated and I lot of companies I see or a lot of CRM companies they still are building for managers. So it’s like report heavy, analytics heavy as if like they are building for this big brothers who are watching. So that was another unlock. The third unlock was and everybody said that I was so wrong about this that we should not go for companies who are already using CRMs. Like frankly speaking, CRM market is like really big. So even if you get like a new customers who are not using a CRM today, you can still get to like 5 million AR, maybe 10 million AR. But to become a hundred billion dollar company, to beat Salesforce, I think, I know it’s hard, but we should go for switching platform. That really paid us, paid off in term, in terms of like understanding what exactly people hate about Salesforce because everybody ends, people hate Salesforce, but why they end up choosing Salesforce, right? There’s a reason for that. Those three things really helped us.
STEVE: It was, there used to be a saying which was no one ever got fired for buying IBM. That was back in the 80s and 90s. And the new one is Salesforce, obviously. So the other question I had for you talking, speaking about companies that don’t have CRM. So they’ve got a blank slate. You mentioned earlier that you could maybe put some JavaScript on their website and you can start to collect website visitors to to seed the database. Is that, is that one approach to getting going if you’ve got no CRM?
ONE: Yes, that’s one approach. Frankly, that was first approach because we were building for early stage startups. So they care more about prospecting. They care more about getting new customers. A little later, they’re like, how can I maintain this customer data where I can upsell or resell or something like that, right? Now what we do, we just push for one-click like Gmail login. You just go, log in with Gmail, it will immediately go through all your contacts and like account and just populate the account objects, the contact objects, and LLM, which is this part is harder where LLM will go through every single email and try to find deals that you had some level of discussion with you already. And it will tell you like, hey, we found 75 deals in your Gmail, 2000 contacts and maybe like 800 accounts. So one-click approve, done. So normally a CRM starting process would take like months or hours or weeks. It takes now like 10 minutes at most. Most it take like two or four minutes. That is how we do things now.
STEVE: And then how about any sensitive communications? How’s that handled where you might have some email threads you don’t want to get harvested?
ONE: Yes, so first of all, we don’t train on your data. That’s the number one. The number two is we all already have like a lot of compliance like HIPAA, GDPR and SOC 2. So that gives those give people some level of like certainty. Okay, they know what they’re doing. And the third thing, we actually hide the email body by default because in a lot of organization internal emails they don’t want to share, but they are okay if you see there was some level of email communication that is logged. So you, they are okay to show you the date, but they’re not okay to show you the subject line or the body. So we by default blur them or like hide them. So if another user wants access internally, they can request by clicking who is the owner of that email. So only the owner of the email can see the emails. The other people cannot. So yeah, that is and that’s like a default settings. You can change it, of course. So that’s how we do.
STEVE: Okay, great. And my final question is other than appearing on podcast episodes and and SEO, what are you doing to create awareness? I’m assuming you weren’t doing guerrilla marketing at Dreamforce this week, but what are the things you do do?
ONE: So when I started building Octolane, it was a lonely journey. Like it’s just like nobody, I didn’t know what I was doing. Frankly, I still don’t know what I’m doing. So what I would do, we would build in public, everything. So every day I would post something about Octolane, like a demo, things that did work, things that didn’t work. It was really transparent and the struggle, the success and every, frankly it was a lot of like failure. There’s not much of success. So I had like four followers, two of them probably like other fake accounts. Now we have like 4,000 followers on Twitter. That’s my personal account. LinkedIn also grew a lot. So we got most of our early customers through building in public in Twitter and LinkedIn. Later I started a newsletter named Coffee with One where I also shared like a monthly update of Octolane, things that we learned and everything. So that is how we got a lot of our early customers and investors because they are like, this is like a live change log. Like they can see, okay, what whatever is maybe it’s working, maybe it’s not, but they are like doing some level of progress every single day. That gave us like a huge upside. Now right now what we do, I post, I still post, we are still like getting, it’s mostly inbound. I don’t do cold outbound. People like this is a huge market. People just hate CRM. Like it’s just like, I don’t know. So it’s just people just keep coming up. We are not ramping up in terms of like trying to do grow as as fast as possible. I know a lot of AI startups are doing that thing, but we’re not focused on revenue. The only metric we care about how much time we save and then retention. If you’re not using our product at least three or four times a week, something is wrong with us. So that’s what we are currently doing. So it’s mostly inbound driven.
STEVE: Great.
SAM: Yeah. All right, well, I have a couple of comments. First of all, I just want to say that the idea of editing data to test to see if it’s bad is brilliant. I I’m already thinking I’m building a KPI around that because that that’s frankly a brilliant idea. And then it reminds me of a company that was early 2000s that they would monitor all the transactions that were happening in the company, and before it would then report it back to you to try to build automated workflows. And I’m thinking with the use of AI, that’s something that would be trivial now to do. You could certainly build LLM to monitor who’s putting in what transactions and actually automate and make suggestions for automation or actually make the automation. So I thought that was clever. And then the last thing, the last comment is, I work with a lot of financial service companies and they’re terrified of AI because they don’t know where their data is going and how it’s going to be processed. I so they’re the the trend in there is for for them to have their own private LLMs. Is that something you’re thinking about doing or already doing?
ONE: What we have seen that they said they want their private LLM, but it just doesn’t work. Things just doesn’t move and one thing that we have done is that, okay, you want to bring your private LLM, that’s totally fine. You can put your API key, but you have to pay like $300,000 more or something like that. Then it would never go through or goes through. Like if it’s important enough, they would pay. We cannot do that yet. You can still fine tune and everything. One thing that work that they have their own contract with OpenAI or like Anthropic and they bring their API key from OpenAI and Anthropic, then it works.
STEVE: All right One, we’ll really appreciate your time today. This is, this is fascinating and wish you the best of luck as you continue to go to market with this very innovative CRM solution.
SAM: Yeah, really good job, very clever.
ONE: Thank you, Sam. Thank you, Steve. Really appreciate everything. And really happy to be here with, I had a great conversation with you guys.