Q&A with Qin Zi’ang, Founder of Capmind: From 'Recording' to 'Execution' — What’s Next for AI Earbuds?

Column Introduction: EBRUN Global Good · Interview is a global archive of standout products built by Matishe Community × Ebrun. We go deep into the frontlines of corporate R&D and manufacturing, dedicated to deconstructing the pioneers redefining user-loved 'great products' amid the wave of Chinese brands going global. Beyond just documenting stories, we aim to review the key decisions and user insights behind every product. Interviewer: Liu Chen (Co-founder of Ebrun, CEO of Matishe Community) Interviewee: Qin Zi’ang (Founder of Capmind) Earbuds are already one of the most saturated categories in consumer electronics. Tech giants compete on audio quality, noise cancellation and ecosystem lock-in; translation earbuds solve cross-language communication; AI recording devices turn conversations into transcripts and summaries. For new entrants, the question is straightforward: what differentiated value can a new pair of earbuds deliver? For Qin Zi’ang, founder of Capmind, the answer lies in embedding earbuds into the post-meeting workflow. Capmind does not aim to build yet another AI tool that 'captures what was said in meetings'. Instead, it focuses on what happens after meetings end: what decisions were made, what commitments were made by whom, what deadlines were set, and whether those action items are actually delivered on. To that end, Capmind starts from real-world work scenarios including meetings, calls and discussions, turns decisions and commitments made in conversations into clear action items, and further links them to responsible owners, deadlines and follow-up processes. If most existing AI meeting tools answer the question 'what was just said', Capmind aims to solve 'what happens next'. This core focus stems from Qin’s long career as an entrepreneur and team leader. Capmind One, the AI earbuds for post-meeting to-do follow-up and task management He wants to fix post-meeting follow-through Ebrun: What drew your attention to the problem of post-meeting execution? Qin Zi’ang: It actually comes from my years of experience as an entrepreneur and manager. Throughout my startup journey, I’ve sat through countless meetings, client calls and team discussions every day. The real pain point for me was never whether meetings were recorded, but that after meetings wrap up, so many action items fall through the cracks. During meetings, things are usually laid out very clearly: who is responsible for what, when it needs to be done, what the next step is. But once everyone leaves the room, tasks end up scattered across WeChat messages, meeting notes, personal to-do lists, or even just people’s memories. By the time we do the weekly review, we often find that it’s not that people lack the ability to get things done — it’s that no one is consistently pushing those tasks forward. That led me to focus on a few very simple questions: who committed to what? Who is the owner? When is it due? Did it actually get done in the end? That’s the original problem Capmind set out to solve. We don’t want to build a better meeting recorder. We want to carry the decisions, commitments and action items formed in meetings and real communications all the way through to execution. In other words, meeting notes are not the finish line — the work that truly matters usually starts right after the meeting ends. Ebrun: Why did this problem resonate so strongly with you personally? Qin Zi’ang: It probably has to do with the fact that I started running projects very early on. I took part-time jobs during school holidays when I was in middle school, and kept trying out different projects after the national college entrance exam and throughout university. Those weren’t formal startups back then, but I already started learning how to identify demand, organize teams, split responsibilities and get things delivered. The summer after I finished the college entrance exam, I ran a college student resource matching program, and tried a tutor matchmaking service. Later in university, I launched training programs, and had to push forward everything from recruiting instructors, enrolling students, designing curriculums to coordinating the team. Those experiences made me realize very early that an idea itself is not worth much — the hard part is turning it into tangible results. Ebrun: Did you experience project failures later on? Qin Zi’ang: I did, and those failures actually had a bigger impact on me. Once, a project performed really well in one regional market, so I tried to scale to other markets quickly the next year — but none of the new markets took off. That was the first time I deeply understood that there is a huge gap between 'figuring an idea out' and 'actually making it happen'. The right direction doesn’t guarantee successful execution; making a plan doesn’t mean things will get done on their own. After graduating from university, I went to Africa to start a business, and after returning to China I kept working in consumer electronics and smart hardware. As my teams grew larger, my role gradually shifted from 'getting things done by myself' to 'getting things done with the team'. That shift made me increasingly sensitive to the topic of execution. Ebrun: How is that directly connected to your decision to build Capmind? Qin Zi’ang: It’s deeply connected. I later found that many small teams don’t lack ideas, nor do they lack tools. What really drains a manager’s energy is that after making dozens of decisions every day, they have to keep following up to check if those decisions are actually being executed. When you work alone, you can rely on your memory and personal habits. But when you lead a team, you start to see information scattered, tasks missed, and context lost. That’s why I came back to the post-meeting problem again. If AI can capture decisions made in natural conversations, and continuously help teams push those tasks forward, it won’t just solve a recording problem — it will solve a real execution problem. That’s where Capmind started. The real problem starts when meetings end Ebrun: You mentioned earlier that Capmind grew out of the management pain points you encountered long-term. What specifically were those pain points? Qin Zi’ang: As an entrepreneur, my biggest problem was never how to run meetings, but how to make sure the arrangements made in meetings actually move forward after they end. As a team leader, I ended up asking the same questions over and over again: Who committed to this? When is it due? What’s the progress right now? Why hasn’t it moved forward? I later realized that what drains managers’ energy is not making decisions, but continuously following through on the decisions that have already been made. Ebrun: So you had this idea a long time ago? Qin Zi’ang: Yes. Back in 2020, when I was leading brand and growth at Timekettle, I tried to build a product in a similar direction. At the time, the pandemic had halted international business travel, and I thought: can we build a device that records meetings and conversations, transcribes them, and helps me keep track of follow-up tasks? We got to the demo stage, but since the company’s core business was translation earbuds, we didn’t continue investing in that direction. Looking back now, that attempt planted the earliest seed for Capmind. It’s just that AI capabilities back then were not mature enough to move from 'recording' to 'understanding and execution'. Ebrun: Beyond your personal experience, have you looked into real-world scenarios of other small and medium business owners? Qin Zi’ang: Later I talked to entrepreneurs and team leaders across different markets about how they work, and found that everyone faces very similar problems. I met a Japanese founder of a PR firm in Southeast Asia, who leads a team of 40 to 50 people. He stands at the front every day taking client requests, running meetings, assigning tasks. He tried Google Tasks, Microsoft To Do, Lark, plus large model tools like ChatGPT and Claude — but ended up tracking most things in Google Sheets. He asked me a question: 'AI is so powerful now, why is there still no product that works well for such a simple need?' That confirmed for me that the problem is not a lack of task management tools on the market, nor that large models are not smart enough. It’s that there is a missing link between the 'decisions' made in meetings and the 'execution' that follows. If the system can understand those communications, identify tasks, responsible owners and deadlines, and keep track of whether those tasks get completed later, managers won’t have to keep moving information back and forth between people and tools. That’s the real problem Capmind wanted to solve: not to build another task management app, but to carry decisions and commitments from communications all the way through to execution. Alex Qin Zi’ang, Founder of Capmind Advances in AI make 'from understanding to execution' possible Ebrun: Why do you think now is the right time to actually build this? Qin Zi’ang: One key change is that AI is evolving from 'understanding information' to 'executing tasks'. The development of MCP (Model Context Protocol), AI agents, and tool-calling capabilities means AI can now connect to calendars, emails, CRM systems and task management tools. Meeting content is no longer just summarized — it can be processed to identify tasks, owners and deadlines, and fed into downstream workflows. For me, the more important significance of MCP is that it makes the full chain from understanding to execution viable. Ebrun: Is that the biggest difference between Capmind and traditional meeting products? Qin Zi’ang: Yes. Many AI meeting tools today do an excellent job at recording, transcribing and summarizing, and those capabilities are foundational for Capmind too, but we don’t want to stop there. What we really want to solve is how to carry decisions and commitments from conversations through to execution. Who committed to what, when it is due, and whether it actually got done. The end of a meeting is not the end of work — it’s where Capmind starts delivering value. Capmind One uses earbuds as the entry point for real-world communication What can startups do after tech giants enter the space? Ebrun: You lived through the cutthroat competition in the translation earbud market. Now entering the AI earbud track again, facing more and more tech giants and established brands entering the space, are you worried? Qin Zi’ang: After going through the last round of the translation earbud market, I’m actually clearer on one thing: big companies have more resources, but resource advantages don’t guarantee they will win every niche scenario. When I attended the Global Sources exhibition in Hong Kong in 2018 and 2019, 30% of exhibitors were making translation earbuds. iFlytek, Sogou, NetEase, Cheetah Mobile, Baidu — all of them made translation earbuds. There were products built by Tsinghua PhDs, by co-founders of Mobike, by Japanese listed companies, and Waverly Labs which raised $4.2 million on crowdfunding… plus all kinds of domestic solution providers and factories, there were an estimated hundreds of players making translation earbuds at the time. But by 2020, some projects had exited; by 2022, even fewer brands remained in the market long-term. That experience made me realize that large companies and startups have different advantages. Big companies have capital, channels and ecosystems, but for a project to keep getting resources, it has to align with the company’s overall strategy and return requirements. Startups have limited resources, but they can focus deeply on a very specific problem, and build deeper user and scenario expertise. So startups don’t need to compete head-on in areas where big companies are strongest. What’s more important is finding your own user base and 'home turf'. Ebrun: But AI earbuds have become a hot track, with products like OpenFit AI and Plaud entering the market one after another. How do you determine that Capmind still has a chance? Qin Zi’ang: I think the key is not 'how many people are making AI earbuds', but what problem everyone is trying to solve. Some focus on high-quality recording, some build personal memory tools, some want to be a general-purpose AI assistant. Capmind is taking a different path: we focus on startup founders and senior executives, with an emphasis on post-meeting task tracking and execution follow-up. So what I care about is not whether the product form is the same as others, but whether we can define this scenario clearly enough, and go deep enough in it over the long term. A crowded track doesn’t mean there are no opportunities. What really determines whether a product can stick around is whether it solves a specific, persistent problem. The integrated hardware-software form of Capmind One Why Capmind is starting with small and medium team managers Ebrun: Who is Capmind’s core user? Qin Zi’ang: At this stage, our primary focus is entrepreneurs, small and medium business owners, and managers who are directly accountable for business results. A common trait of this group is that they both make decisions and personally push execution forward. In large companies, work usually follows relatively mature processes and clear role divisions; but in a team of a dozen to dozens of people, a manager might discuss product in the morning, meet clients in the afternoon, and follow up on recruiting and operations in the evening. Everyone wears multiple hats, so information and tasks are easily scattered across meetings, chat histories and personal to-do lists. For them, the real time drain is not a 'lack of tools', but having to constantly confirm what decisions have been made, who is responsible, and what the current progress is. Ebrun: Is this a problem you encountered yourself as a manager? Qin Zi’ang: Very much so. For example, after a morning meeting with the product and testing teams where we finalized several test items, when I discuss hardware in the afternoon, I might already not remember all the details from the morning. It’s not that those things are unimportant, it’s that there are too many contexts to handle in a day. Many small teams are the same. Decisions are made in meetings, but after the meeting tasks end up scattered across chat logs, meeting notes, and people’s heads. By the time the next weekly meeting comes around and we realize something didn’t move forward, time has already passed. That’s why I’ve always believed that the value of a meeting is not getting a set of notes, but making sure the decisions made in the meeting are actually pushed forward. Ebrun: Why are you prioritizing small and medium teams at this stage, instead of large enterprises? Qin Zi’ang: It’s not that large enterprises don’t have this need, but the pain point is usually more acute for small and medium teams. Large enterprises often have project managers, process systems and clearer role divisions; small and medium teams prioritize speed, a single person might wear multiple hats, and it’s hard to assign dedicated staff to record, organize and follow up on every conversation. So they need a system that can proactively understand from daily communications: what decisions were made, who is responsible, when things are due, and whether progress is being made. That’s why we chose to start with this group of users. It’s not that small and medium teams are poorly managed, it’s that they have fewer people handling a higher density of decisions and collaboration. We want to go deep in this scenario first, then gradually serve more types of teams and organizations. Capmind One AI communication scenario Put tasks front and center: be the boss’s Execution Copilot Ebrun: What is the core problem Capmind wants to solve right now? Qin Zi’ang: At this stage, what we want to nail completely is post-meeting and post-conversation task follow-up. Recording, transcribing and summarizing are important, but they are foundational capabilities. If a meeting only ends up generating a set of notes, but doesn’t help the team reduce missed items and push work forward, its value is limited. So we focus more on: what decisions were made in the meeting, who is responsible, when they are due, and whether those things keep moving forward after the meeting. Ebrun: So the difference between Capmind and traditional recording-transcription products is not just the user base? Qin Zi’ang: Right, the user base is just our entry point. The real difference is the problem the product solves. Capmind will also do recording, transcription and summarization well, but we don’t see 'transcription complete' as the end point. We want to go further to identify post-meeting tasks, responsible owners and next steps, and use reminders and follow-ups to turn those decisions into actual execution. For startup founders and senior executives, what drains the most energy is often not the meeting itself, but the constant post-meeting check-ins: Who is doing this? What’s the progress? Did anything fall through the cracks? We want to hand that repetitive management work over to AI. Ebrun: How do you define Capmind’s role in a team? Qin Zi’ang: It’s not a secretary — a secretary feels like someone who does everything for you. It can’t just be an assistant either, since there are levels of assistants. I prefer to think of it as an Execution Copilot. Traditional project management tools usually require users to manually enter tasks, but a lot of real work doesn’t start by opening a task app. Tasks often come up in a meeting, a phone call, or a quick impromptu discussion. Capmind aims to first understand these naturally occurring conversations, capture the decisions and action items that come out of them, then help managers keep visibility on subsequent execution status. Ebrun: Looking beyond post-meeting task follow-up, where will Capmind go next? Qin Zi’ang: We will first go deep into the post-meeting task follow-up scenario, which is our most important 'home base' right now. Further down the line, as AI develops a more complete understanding of team context, it can gradually expand from individual tasks to project collaboration, helping managers understand progress, risks and priorities across different projects. But no matter where the product goes in the future, the core won’t change: we don’t generate more information — we help teams actually get done the things they have already decided to do. Capmind AI meeting task follow-up Why use earbuds to carry task workflows? Ebrun: What is Capmind’s uniqueness and technical moat on the hardware side? Qin Zi’ang: Our core capability is in software; hardware is the entry point. Today’s AI hardware comes in form factors including pendants, wristbands, watches, employee badges, rings, glasses, and earbuds. What is the device users can’t live without? If we count smartphones as a portable device, the first thing users can’t live without is their phone. But among these new AI hardware form factors, I believe earbuds are the easiest to integrate into existing usage habits. Why? Because fewer and fewer people hold their phones to their ears to take calls now — most people wear earbuds to take calls, join meetings, and listen to music. Earbuds are already a very mature device form factor that users barely need to learn how to use. What’s more, a lot of AI hardware today emphasizes 'perception', but I think perception alone is not enough — you also need interaction. Earbuds are naturally built with microphones and speakers, so they can both receive information and deliver reminders and feedback to users. That’s another reason I think earbuds are better suited for this use case. I sum it up in one phrase: uphold the fundamentals while pursuing innovative breakthroughs. The 'fundamental' part is not changing users’ basic habits, letting the product fit naturally into existing behaviors; the 'breakthrough' part is using AI to turn earbuds from a tool that just records information into something that can understand conversations and participate in downstream work. Recording is just the first step. More importantly, earbuds can both receive information and interact with users. For example, after a meeting ends, the system not only knows 'what was just discussed', it can also tell you which items need follow-up. So I believe truly valuable AI hardware can’t just 'perceive' — it has to be able to form continuous interaction with users. Earbuds handle getting into conversations, but the real product value happens after the conversation ends. Ebrun: So the focus of hardware is not just audio quality and specs? Qin Zi’ang: Right. Specs are important, of course, but they ultimately need to serve real scenarios. Right now Capmind aims to cover five high-frequency scenarios: phone calls, online meetings, offline meetings, on-the-go communication, and media content. For founders and senior executives, important work information doesn’t only appear in conference rooms or in front of computers. It might come from a client call, a Zoom meeting, an in-person discussion, or even during a commute or impromptu chat. The value of earbuds is that they can naturally span all these scenarios, and connect originally scattered communication threads. So instead of just piling on specs, we care more about whether users can wear the earbuds long-term, use them stably, capture important information across different communication scenarios, and feed that information into downstream task and execution workflows. Capmind AI role-specific meeting to-do progress dashboard Starting from overseas markets to validate real product value first Ebrun: Will Capmind prioritize the Chinese market or overseas markets at this stage? Qin Zi’ang: We will prioritize overseas markets for now. My past experience has made me relatively familiar with overseas users and markets, but we won’t treat 'overseas' as a single homogeneous market. We will gradually select suitable markets for validation based on user demand, competitive landscape, channel costs and product acceptance. Compared to mature large companies, we need to first find a 'home base' where we can quickly validate product value, then expand outward. For Capmind, what we care most about is not how big a market looks on paper, but: is there a group of users who actually have this problem, and are willing to keep using the product and pay for it to solve their problem? Ebrun: Have you seen clear user feedback so far? Qin Zi’ang: Yes. During the product testing phase, we worked with a CEO of an investment consulting firm from Malaysia, who has a huge number of meetings every day — a very typical high-frequency communication and team management user. After using Capmind, he used the information captured from meetings for client consulting and follow-up work, and was willing to pay for the product on an ongoing basis. For us, that kind of feedback matters more than raw download numbers or trial sign-ups. Because it validates one thing: whether users will actually integrate Capmind into their workflow, and rely on it consistently to solve problems. That’s what we most want to validate in overseas markets at this stage. Ebrun: What are your thoughts on fundraising? Qin Zi’ang: We are open to fundraising, but we don’t see fundraising itself as a goal. If we meet investors who truly understand this mission, can provide long-term value, or create synergies in areas like market access, supply chain, or channels, we are absolutely open to partnering with them. But at the same time, we want the company itself to have the ability to generate sustainable revenue. I’ve always believed that funding is an accelerator, not a prerequisite for a product to work. The fundamental logic of a product company ultimately comes back to users: are people willing to keep using it, are they willing to pay for it continuously, and is the product consistently creating real value? So instead of focusing on 'how much money we’ve raised', right now we care more about: can we find stable user demand as quickly as possible, validate our product value, and build a healthy business loop. Capmind overseas user product experience We don’t just need executors — we need people who can define problems Ebrun: The early team’s product didn’t fully meet your expectations. What was the main problem? Qin Zi’ang: I later realized that it was less a problem of ability, and more a mismatch between ability and role. Someone might be excellent in their original product domain, but when moving to AI earbuds, they have to face a whole set of cross-domain issues including acoustics, wear comfort, hardware experience, and AI interaction, which is not as easy as it sounds. That made me increasingly clear: early startup teams don’t need people who are exceptionally strong in one single skill, but people who can understand problems, learn fast, and get things done across product and technology boundaries. Ebrun: What kind of people are you looking for most right now? Qin Zi’ang: Internally we call this role a Product Engineer Manager. It’s not a traditional pure engineer, nor a product manager who only handles requirements and processes. This person needs to understand technology, product and users at the same time. They need to know what AI can do, and also understand why users need a feature, and which problems are higher priority to solve. On the technical side, they need to understand the capability boundaries of agents, skills and AI systems; on the product side, they need to be able to tell real demand apart from features that just look cool. Simply put, we want to find people who can both define problems and drive solutions. It’s the same for the marketing team. AI tools can improve efficiency, but what we value more is an understanding of the market, brand and users — the ability to clearly articulate who the users are, what their pain points are, and why our product is worth choosing. Ebrun: But people like that must be in high demand in the job market. Why would they choose Capmind? Qin Zi’ang: Because this product category hasn’t been fully defined yet. Mature large companies usually offer people a relatively clearly defined role, but at Capmind, many problems still need to be defined from scratch. What kind of AI earbuds do users actually need? After meeting recording, what else can AI do? How should humans interact with AI? None of these answers are fully settled yet. So what we can offer people is not just a job, but an opportunity to participate in defining a product, even a new category, from zero. Many decisions they make today will directly shape what Capmind becomes in the future. That kind of ownership is one of the most important draws of an early-stage startup. Qin Zi’ang didn’t enter the AI earbud track by first picking a new hardware form factor and then looking for use cases for it. On the contrary, he first picked a problem he had lived with and encountered repeatedly for years: for entrepreneurs and team managers, how to make sure decisions made in meetings are actually pushed forward and completed after meetings end. That’s also why Capmind is starting with small and medium business owners and senior executives at this stage. This group often wears multiple hats, constantly making judgments, assigning tasks and pushing projects amid high-frequency communication, but can’t rely on the complete processes and dedicated roles that large companies have to carry out every decision. For that reason, Capmind doesn’t want to be just another AI recording earbud. Recording, transcription and summarization are just the starting point. It wants to answer the post-meeting questions: who committed to what, when it is due, and whether things are actually being pushed forward. Behind that is Capmind’s longer-term vision — Extend Human Agency. AI shouldn’t just help people remember more information; it should help people keep pushing forward the decisions they have already made. Of course, whether this direction will ultimately work still needs to be proven by real users: whether task recognition is accurate, whether post-meeting follow-up actually reduces missed items, whether users are willing to keep using and paying for the product — all of these questions will be answered after the product launches to the market. But for an emerging category still taking shape, the opportunity might lie exactly here: not looking for a track with no players, but finding a problem that hasn’t been truly solved yet in an already crowded market, and going deep on it persistently. This article was first published on the official website of Ebrun.