Quitting his job and growing an AI product to $3M ARR in 12 months

Frank Sondors quit a cushy sales job to build an AI product called Salesforge. Within 12 months, he turned it into a $3M/yr product ecosystem. Here's Frank on how he did it. 👇 Contents Build what you know Time is money Software compound model Integrated marketing Biggest failures Talk to your customers What's next? Build what you know I led a sales team of 50 as VP of Sales at a large company and realized that the only way to scale a sales-led organization was by having a bigger headcount. But that didn't seem like the right way to scale a company. So eventually, I left, and I had two choices: Either I would become the head of sales for another company with a sweet six-figure compensation, or I'd start my own company. I felt like I would just be rinse-and-repeat when it came to being the head of sales and selling software for another CEO. I thrive where I'm uncomfortable, as do the best salespeople, so starting my own company and then burning through my savings seemed like the best path forward. I founded Salesforge, which helps B2B companies to blend humans and AI agents to maximize sales pipelines and reduce the cost of customer acquisition (CAC). We've scaled from to $3M ARR in the first 12 months, which made us the fastest growing company in our category. And In less than 2 years we've built 6 products as part of the software compound model. Salesforge.ai: Executes sales across email and soon LinkedIn with humans and AI agents. AgentFrank.ai: AI SDR that lives within Salesforge and is powered by all the forges. Infrաforge.ai: Private cold email infrastructure providing domains and mailboxes Mailforge.ai: Distributed cold email infrastructure providing domains and mailboxes Warmforge.ai: Email deliverability command center to boost your email infrastructure health Leadsforge.ai: Launching in March, ChatGPT-like interface finding leads for your ICP Time is money We didn't have much time — remember, time is money. I was working full time on the forge while my cofounders were in full-time jobs. I was essentially burning through all my savings. And I have two kids that I also had to support, so it was definitely tough. We didn't have any outside funding at this point. We didn't have an office. We actually worked from an IKEA co-working space in a mall. I never imagined that IKEA would ever offer co-working space. Prior to actually shipping the product, I did 40 user interviews with different potential buyer personas from different types of companies and geographies. And then, before shipping a single line of code, I managed to get four companies to pay for the product. And they're still customers to date. That's how I validated the product. Also, as I'm an expert in this domain after over a decade in sales, I was able to accelerate the validation as I'm building the product for my former self. In the end, we were able to ship V1 of the product within 2-3 months. Here's our stack: Frontend: React + Typescript Backend: Golang, Python Hosting: AWS Software compound model Similar to Rippling, we're following the software compound model approach, which means we're selling multiple software products, or what we call "forges", to the same users and the same buyers, which significantly improves our unit economics. Every product that we sell, we charge for consumption and we operate a land-and-expand strategy. We charge very, very low for the product and, as the user sees the value in the product, they increase their consumption — sometimes 1,000X+ within months. That's vertical expansion, but then there's horizontal expansion, where we want to sell additional forges to customers. We currently have five products and are about to launch leadsforge.ai, our final forge, which completes the Forge Ecosystem V1. Think of it like Apple or Microsoft, where you have a suite of products. In the next 18-24 months we'll build at least another three forges to improve pipeline numbers. The biggest advice I have is to figure out how to sell multiple products a) direct, b) to other tech companies in the space (so take one of the forges and power other parties in the space), and then also unlock c) the reseller model, similar to GoHighlevel, where the majority of the revenue comes from your unofficial sales army via revenue share — be that consultants, agencies etc. As a company, we were always profitable, even though we did raise $500K as pre-seed. The reason we raised was that we thought we would burn cash, but that just never happened, so that just sits in the bank, which is a good problem to have. Integrated marketing From day zero we've tried to be everywhere because we know that if our users see us in multiple places, the probability of a conversion will be much, much higher. It's called an "integrated marketing approach". It was also about doing things that don't scale: Slack communities via Hiveindex and introducing myself to everyone. I would scrape Slack communities using Phantombuster. Dogfooding Salesforge and Infraforge, and sending a ton of cold emails. Posting on Reddit about ourselves and commenting on posts of where competitor's keywords mentioned using F5bot. Offline events — I managed to attend a ton of offline events. Messaging people in my network like crazy. Product Hunt, naturally. Not for sales, but more for buzz and getting the backlink. A ton of videos on YouTube in the early days. And one of the best decisions I've made is by building in public and sharing my journey on LinkedIn and showcasing how we're solving pains for our customers. I share all the high and lows of the journey. I make sure that I post once per day Monday through to Friday and sometimes even on weekends. Biggest failures We've failed plenty of times and I think the biggest failures would be specifically around people, which is something that's very, very typical in every startup. You simply can't find the right people for every business, so you need to learn to hire fast, but at the same time you need to be even better at firing fast. On the product front the biggest errors were scoping errors. Essentially there's a huge difference between how I imagine the product should work, being the domain expert, and my cofounders imagine the product should work. We've failed multiple times on this. We're building world-class software and, hence, similar to Steve Jobs, I'm very particular about the UI/UX of the software ecosystem that we're building. Talk to your customers It should be obvious, but folks just don't do it: Talk to your customers. I always say people should install a chat widget on your website from something like Intercom or Gleap so that you give the user the ability to ask you any question in real time. So many folks distance themselves from their users, but you literally need to become available to everyone across multiple platforms. The other thing that I would highly recommend is to leverage n8n or make.com to automate most of the repetitive tasks in your business, so you don't have a pressure on hiring people to do various jobs. The last thing would be to feed your brain with knowledge from others, and I typically do this by joining two networking platforms, which don't cost you a dime. One is lunchclub.com and the 2nd one is saleshookup.com. Apart from these two platforms, I engineer my feed where I follow people like Jason Lemkin, Nathan Latka, cool startup founders from companies like Chili Piper, Clay, and even upcoming ones like Face2face.io. What's next? We're looking to roll out three new products this year. We also want to 3.3x our revenue to truly become a software ecosystem of best-in-class tools helping companies build sales pipelines with the least possible headcount. The biggest challenge will be navigating the red ocean of 100+ competitors. You can follow along on X, LinkedIn, and YouTube. And check out our ecosystem: salesforge.ai leadsforge.ai agentfrank.ai infrаforge.ai mailforge.ai warmforge.ai

Frank Sondors quit a cushy sales job to build an AI product called Salesforge. Within 12 months, he turned it into a $3M/yr product ecosystem.