#045 - Kaushal Subedi & Ashish Ghimire - From Vague to Focused: How Echowin Found ICP Clarity Through Rapid Positioning Iteration
Wed Jan 14 2026
The foundation of this episode rests on a critical challenge that most early-stage B2B SaaS companies face but few talk about honestly: how do you find positioning clarity when your market is being invented in real time? Echowin launched in November 2022, before ChatGPT was public. At that point, voice AI was still using keyword-based natural language processing—the "press one for billing, press two for sales" systems we all hate. Kaushal had early access to GPT-3 and built a voice assistant on his Apple Watch. The lightbulb moment came when he watched his mom, a small business owner, drop everything to answer a phone call while serving a client.
Kaushal describes the moment: "She was with her client randomly gets a phone call, she has to drop everything she's doing, run to the phone, and she was speaking in a hurry with a person on the other side of the phone call. There was a lot of tension building up. I could see the client that was waiting, like they were clearly like, 'What's going on?' My mom was speaking in a rush. I'm pretty sure the person on the other side of the phone call felt that too. That's when it all kind of clicked."
Within seven days, Kaushal quit his job at Amazon Robotics. Ashish quit his aerospace job. Within 15 days they had a working prototype. Within three months, paying customers. But having a product and having positioning clarity are two very different things. And that's where the real journey begins.
Early positioning was broad—too broad. Kaushal admits: "Our positioning was something that we were still figuring out. It was very wide and very vague." They started as a "full horizontal platform" targeting all small businesses. The messaging emphasized value props like "no missed calls" and "natural language understanding," but the ICP was unclear. There are 35 million small businesses in the US alone. Who exactly were they for? The answer: they didn't know yet. They were gathering signal, running experiments, and watching what stuck.
The expensive lesson: when messaging attracts the wrong crowd. One of the first major pivots came when they tested the message: "Build your AI agent in less than 5 minutes." The goal was to emphasize speed and ease. The result? It attracted the wrong crowd. Ashish describes it: "That messaging drew the wrong crowd and got us in a lot of trouble because the mass market started coming in with wrong expectations. They didn't understand the limitations of technology and we were unable to explain that clearly. And people would come in, pay for the platform and they would churn. It was an expensive lesson."
The mismatch between promise and reality created friction, frustration, and churn. So they adjusted. The next iteration was: "Build your AI agent." Not in 5 minutes. Just build. This subtle shift changed everything. Ashish explains: "We started saying 'Build your AI agent.' We started attracting the builder persona, these early adopters, slightly semi tech-savvy people who wanna tinker and build things out. That's how the platform evolved from such and such platform to a builder platform where we were naturally attracting builders."
The breakthrough came from cohort analysis. Kaushal and Ashish went back through their customer data and asked: who's been here for a year? Who built something on their own? Who's generating high call volumes? The pattern was clear. Ashish describes the insight: "We went back, we did extensive cohort analysis of who was getting benefit out of the platform. We looked at our existing customers, the customers who got excited, who built it on their own. In some scenarios, we even offered help and they're like, 'Nah, I got this.' We are seeing success again and again and again with this persona."
That insight allowed them to refine their messaging, narrow their ICP, and speak directly to solution-aware buyers. The current Echowin homepage reflects this clarity. Kaushal explains: "Now that we have a much clearer idea of who we're targeting and who the messaging is for, we can already assume some things about them. They know what agents are, they know what these things do. When these builders come to our platform, they're not looking for the high level of what these agents can do. Instead they're looking for, why should I pick this platform over all the other ones out there?"
This is a critical distinction. Early on, Echowin had to educate prospects on the category—what voice AI could do, why it mattered, how it was different from old IVR systems. Now, they're speaking to people who already understand the space and are evaluating platforms. That shift from problem-aware to solution-aware messaging is one of the most important transitions a B2B SaaS company can make. And it only happens when you know your ICP deeply.
Training humans to talk to AI. One of the most interesting insights from this conversation comes from Ashish, who describes a cultural challenge they're facing: "One of the interesting things that we've seen is we are in the process of training these agents. At the same time, I feel like sometimes we're training humans to talk to AI too, because over the course of the last 20 years, we've been trained to interact with these bots—press one for this, two for that. Now, anytime people hear anything robotic, they're smashing their zero on their phone."
He continues: "Over the course of the last two and a half years, people wouldn't even interact with these agents. They had no idea you can actually have a fluent conversation and they can actually help you. But things are changing now. We have seen drastic difference."
The AI hype cycle and messaging fatigue. As AI became oversaturated in 2023 and 2024, Echowin had to adapt their messaging again. Kaushal describes the challenge: "It even got to a point where the hype cycle with AI was in full motion and people were just dismissing things without even fully diving into what the platform does. Just because everyone was throwing the word AI but not delivering properly. So instead of that, we wanted to emphasize what the platform does—in this case, answers calls."
This is a tactical lesson: when a category term becomes noise, shift focus to the outcome. Don't lead with "AI-powered." Lead with "answers your calls 24/7." The technology is the enabler, not the value prop.
How they use data and customer conversations to drive messaging. Both founders emphasize being data-driven but adaptive. They use PostHog and Google Analytics to track conversion funnels, run A/B tests on headlines and CTAs, and measure which messaging drives signups. But they also stress the importance of talking to customers.
Kaushal shares a tactical example: "We use Fireflies internally. We take the transcripts from our Fireflies calls with some of the agencies and we're like, 'Hey, Claude, find me all the top questions here.' And then we use that as information as well." Those questions inform their FAQ section, their homepage copy, and their sales collateral.
Ashish adds: "Nothing beats interacting with customers. You would see their eyes spark, honestly. They'd be like, 'My goodness, Echowin is such and such on steroids.' You know, they had been testing all these different tools and with what we offered and what they could build with it, they were clearly super excited."
Watch for the moment their eyes widen. Kaushal shares what to look for in customer conversations: "Usually people will either say, 'Oh yeah, that part makes sense, that part makes sense.' And then there'll be some part where their eyes will widen up and they'll be hooked onto their screen. You wanna keep, take note of that, those parts and then you work backwards from that."
From drag-and-drop to document-based training. One of Echowin's biggest product and messaging pivots was simplifying how users train AI agents. Early on, they used a drag-and-drop builder (like n8n or Zapier workflows). You'd create scenarios: "When a caller asks for a refund, ask this question, then send an email." It worked for demos, but it didn't scale for real-world conversations. Real conversations are dynamic. People jump between topics.
So Echowin rebuilt the platform around a document-based training system. Kaushal explains: "The experience we were trying to replicate is exactly this: If you are hiring a receptionist or someone in your call center, you are probably gonna give them some instructions, whether that's an email or standard operating procedure document or something. We just want you to be able to copy paste that here just once."
This simplification was key to attracting non-technical builders and reducing time-to-deployment. Everyone knows how to use Google Docs. Now everyone can train an AI agent.
Don't get hyperfocused on competitors. Kaushal's advice on competitive positioning is refreshingly honest: "First, I always tell my team, 'Hey, don't be hyperfocused on what the competition does,' just because first this is a very new industry, emerging market, emerging technology, meaning nobody has their shit together. Everyone's figuring out which path to go, how to position, how to message. Sometimes when you get too hyperfocused on competition, you start focusing on the wrong things."
He continues: "We see competitors taking inspiration from some things we do and we see them experimenting and sometimes some of the things we see on their landing page or wherever it might be, that we think is, 'Oh, this is amazing that they did that,' just goes away in a week and then we're like, 'Okay, I guess that didn't work out.'"
The lesson: focus on your customers, not your competitors. Kaushal emphasizes: "The best source of truth we have right now is our interactions with customers. They're the ones telling us, 'Hey, this is working, this is not working. This is the type of information I want. This is the type of information I already know.' And we wanna work backwards from that, not backwards from what someone else is doing."
Product iteration should
More
The foundation of this episode rests on a critical challenge that most early-stage B2B SaaS companies face but few talk about honestly: how do you find positioning clarity when your market is being invented in real time? Echowin launched in November 2022, before ChatGPT was public. At that point, voice AI was still using keyword-based natural language processing—the "press one for billing, press two for sales" systems we all hate. Kaushal had early access to GPT-3 and built a voice assistant on his Apple Watch. The lightbulb moment came when he watched his mom, a small business owner, drop everything to answer a phone call while serving a client. Kaushal describes the moment: "She was with her client randomly gets a phone call, she has to drop everything she's doing, run to the phone, and she was speaking in a hurry with a person on the other side of the phone call. There was a lot of tension building up. I could see the client that was waiting, like they were clearly like, 'What's going on?' My mom was speaking in a rush. I'm pretty sure the person on the other side of the phone call felt that too. That's when it all kind of clicked." Within seven days, Kaushal quit his job at Amazon Robotics. Ashish quit his aerospace job. Within 15 days they had a working prototype. Within three months, paying customers. But having a product and having positioning clarity are two very different things. And that's where the real journey begins. Early positioning was broad—too broad. Kaushal admits: "Our positioning was something that we were still figuring out. It was very wide and very vague." They started as a "full horizontal platform" targeting all small businesses. The messaging emphasized value props like "no missed calls" and "natural language understanding," but the ICP was unclear. There are 35 million small businesses in the US alone. Who exactly were they for? The answer: they didn't know yet. They were gathering signal, running experiments, and watching what stuck. The expensive lesson: when messaging attracts the wrong crowd. One of the first major pivots came when they tested the message: "Build your AI agent in less than 5 minutes." The goal was to emphasize speed and ease. The result? It attracted the wrong crowd. Ashish describes it: "That messaging drew the wrong crowd and got us in a lot of trouble because the mass market started coming in with wrong expectations. They didn't understand the limitations of technology and we were unable to explain that clearly. And people would come in, pay for the platform and they would churn. It was an expensive lesson." The mismatch between promise and reality created friction, frustration, and churn. So they adjusted. The next iteration was: "Build your AI agent." Not in 5 minutes. Just build. This subtle shift changed everything. Ashish explains: "We started saying 'Build your AI agent.' We started attracting the builder persona, these early adopters, slightly semi tech-savvy people who wanna tinker and build things out. That's how the platform evolved from such and such platform to a builder platform where we were naturally attracting builders." The breakthrough came from cohort analysis. Kaushal and Ashish went back through their customer data and asked: who's been here for a year? Who built something on their own? Who's generating high call volumes? The pattern was clear. Ashish describes the insight: "We went back, we did extensive cohort analysis of who was getting benefit out of the platform. We looked at our existing customers, the customers who got excited, who built it on their own. In some scenarios, we even offered help and they're like, 'Nah, I got this.' We are seeing success again and again and again with this persona." That insight allowed them to refine their messaging, narrow their ICP, and speak directly to solution-aware buyers. The current Echowin homepage reflects this clarity. Kaushal explains: "Now that we have a much clearer idea of who we're targeting and who the messaging is for, we can already assume some things about them. They know what agents are, they know what these things do. When these builders come to our platform, they're not looking for the high level of what these agents can do. Instead they're looking for, why should I pick this platform over all the other ones out there?" This is a critical distinction. Early on, Echowin had to educate prospects on the category—what voice AI could do, why it mattered, how it was different from old IVR systems. Now, they're speaking to people who already understand the space and are evaluating platforms. That shift from problem-aware to solution-aware messaging is one of the most important transitions a B2B SaaS company can make. And it only happens when you know your ICP deeply. Training humans to talk to AI. One of the most interesting insights from this conversation comes from Ashish, who describes a cultural challenge they're facing: "One of the interesting things that we've seen is we are in the process of training these agents. At the same time, I feel like sometimes we're training humans to talk to AI too, because over the course of the last 20 years, we've been trained to interact with these bots—press one for this, two for that. Now, anytime people hear anything robotic, they're smashing their zero on their phone." He continues: "Over the course of the last two and a half years, people wouldn't even interact with these agents. They had no idea you can actually have a fluent conversation and they can actually help you. But things are changing now. We have seen drastic difference." The AI hype cycle and messaging fatigue. As AI became oversaturated in 2023 and 2024, Echowin had to adapt their messaging again. Kaushal describes the challenge: "It even got to a point where the hype cycle with AI was in full motion and people were just dismissing things without even fully diving into what the platform does. Just because everyone was throwing the word AI but not delivering properly. So instead of that, we wanted to emphasize what the platform does—in this case, answers calls." This is a tactical lesson: when a category term becomes noise, shift focus to the outcome. Don't lead with "AI-powered." Lead with "answers your calls 24/7." The technology is the enabler, not the value prop. How they use data and customer conversations to drive messaging. Both founders emphasize being data-driven but adaptive. They use PostHog and Google Analytics to track conversion funnels, run A/B tests on headlines and CTAs, and measure which messaging drives signups. But they also stress the importance of talking to customers. Kaushal shares a tactical example: "We use Fireflies internally. We take the transcripts from our Fireflies calls with some of the agencies and we're like, 'Hey, Claude, find me all the top questions here.' And then we use that as information as well." Those questions inform their FAQ section, their homepage copy, and their sales collateral. Ashish adds: "Nothing beats interacting with customers. You would see their eyes spark, honestly. They'd be like, 'My goodness, Echowin is such and such on steroids.' You know, they had been testing all these different tools and with what we offered and what they could build with it, they were clearly super excited." Watch for the moment their eyes widen. Kaushal shares what to look for in customer conversations: "Usually people will either say, 'Oh yeah, that part makes sense, that part makes sense.' And then there'll be some part where their eyes will widen up and they'll be hooked onto their screen. You wanna keep, take note of that, those parts and then you work backwards from that." From drag-and-drop to document-based training. One of Echowin's biggest product and messaging pivots was simplifying how users train AI agents. Early on, they used a drag-and-drop builder (like n8n or Zapier workflows). You'd create scenarios: "When a caller asks for a refund, ask this question, then send an email." It worked for demos, but it didn't scale for real-world conversations. Real conversations are dynamic. People jump between topics. So Echowin rebuilt the platform around a document-based training system. Kaushal explains: "The experience we were trying to replicate is exactly this: If you are hiring a receptionist or someone in your call center, you are probably gonna give them some instructions, whether that's an email or standard operating procedure document or something. We just want you to be able to copy paste that here just once." This simplification was key to attracting non-technical builders and reducing time-to-deployment. Everyone knows how to use Google Docs. Now everyone can train an AI agent. Don't get hyperfocused on competitors. Kaushal's advice on competitive positioning is refreshingly honest: "First, I always tell my team, 'Hey, don't be hyperfocused on what the competition does,' just because first this is a very new industry, emerging market, emerging technology, meaning nobody has their shit together. Everyone's figuring out which path to go, how to position, how to message. Sometimes when you get too hyperfocused on competition, you start focusing on the wrong things." He continues: "We see competitors taking inspiration from some things we do and we see them experimenting and sometimes some of the things we see on their landing page or wherever it might be, that we think is, 'Oh, this is amazing that they did that,' just goes away in a week and then we're like, 'Okay, I guess that didn't work out.'" The lesson: focus on your customers, not your competitors. Kaushal emphasizes: "The best source of truth we have right now is our interactions with customers. They're the ones telling us, 'Hey, this is working, this is not working. This is the type of information I want. This is the type of information I already know.' And we wanna work backwards from that, not backwards from what someone else is doing." Product iteration should