A 5-figure-MRR success after a failed product buried him in debt - Indie Hackers
Join
CompanyKlipy.aiFounderJung Hong KimRevenue>$10K a month
When Jung Hong Kim's product went under during COVID, he went into debt. It took him six years to dig himself out, and along the way, he came up with a new idea.<br>He launched Klipy.ai in three months. And today, it's bringing in a 5-figure MRR.<br>Here's Jung on how he did it. 👇
When a failure propels you toward a new success<br>I am a Korean serial founder. I began my startup journey in Hong Kong. I built and sold two companies developing machine-vision-based retail analytics software for shopping malls and government properties.<br>I was working on another startup — also heavily focused on retail — but it crashed violently when COVID wiped out the entire retail solution market overnight. This put me into serious debt, and I spent about six years as a management consultant specializing in large-scale infrastructure and enterprise architecture.<br>That's when I met my now cofounders. We worked together on projects, witnessing patterns of failure in enterprise software. And that led to the concept of what we are building now. — an AI Chief Revenue Officer that automates all back-office operations for enterprise and consultative sales processes, turning every seller into a 10-person sales team.<br>Most business information system failures stem from the gap between how humans think and how data is saved. Not everyone can think in spreadsheets. LLMs bridge this gap effectively. We eliminate the entire sales data collection process and use AI agents to execute mundane sales processes, allowing sellers to focus on client interactions.<br>The product launched as an automatic CRM in November 2024 and has consistently pivoted and improved. Currently, it serves around 4,000 companies worldwide, mainly in North America and Australia.<br>We're at a 5-figure MRR, and we're targeting $1.5M ARR by the end of 2026.
Building the product<br>This was a classic dogfooding case. I have been a loyal HubSpot customer throughout my career, but I found it painful to enforce its use when scaling sales teams. CRMs are crucial for business operations, enabling forecasting, planning, and centralized customer record-keeping. However, salespeople often lack the motivation to perform data entry. So, we started the product with one feature: a simple sales CRM that automatically logs communications from email, LinkedIn, and meetings using AI.<br>I hand-coded the first MVP in about three months. The stack is Next.js + Convex as the backbone. We have many side systems built with Go and Rust, hosted on Google Cloud, for integration and various subsystems that our AI agents use for scraping and generating documents.<br>Convex helped me save significantly on DevOps costs because it handles the entire infrastructure provisioning and workload orchestration via a JavaScript-based SDK — basically, Supabase for NoSQL. This helped me focus purely on business logic, which accelerated the process.<br>The biggest challenge was the frontend. When I started, I didn't even know what Next.js was. But like any other engineering problem, I built, tested, and set up good observability to identify problems faster than users, then rapidly debugged them.
Landing on freemium<br>We have been bootstrapping this company since day one, with everyone full-time. Fortunately, I am both a developer and a seller. So, we kept our costs very low. We leveraged as many government grants as possible to fund the business, keeping fixed costs as low as possible.<br>Since we were all full-time on this, we had ample time. Money came from grants and our own savings until we became profitable.<br>Our current business model is freemium. We started charging from day one because we didn't want to test or build features based on the feedback of someone who's unwilling to pay. That sort of feedback is mostly nice-to-haves. Burning needs from real customers are more important and should take up 80% of your time.<br>The free tier includes 200 tokens per month, a single user, and two channel integrations. The paid tier ranges from $39 per month per seat up to $149 per month per seat. It varies by monthly tokens (cheaper per token on higher tiers) and optional enterprise security features.<br>We tested many pricing models. These included add-ons such as a monthly fee on channels, token-based pricing, and lifetime deals. Ultimately, we settled on results-based pricing because it makes sealing the deal easier. We then engineer the product to keep margins intact. This also allowed us to navigate significant PLG-driven growth by offering free tokens for specific user actions within the product.<br>We still operate with only the three founders. We are currently raising capital to scale the business, with one pre-seed investor on our cap table.<br>My advice? Charge first, then ensure they feel sufficiently supported. In B2B, people ultimately pay for a sense of security, not features. Features create that...