I failed 4+ startups, then scaled an AI tool to $20K+ in revenue. Here's how.

I wanted to become an indie hacker, work on my own products and generate money with them for the longest time. I tried many different products, but none of them really worked for me. I worked on a single product for almost 7 years and made a grand total of, wait for it, SIX DOLLARS, in all that time.

Then I made MemoryPlugin(.com) and I've crossed more than $20K in revenues in the past year. Here's what changed.

I think one difference was I never really expected MemoryPlugin to go anywhere, I did not expect to actually make a single dollar from it, so I set my expectations to basically nothing. I'd recently taken the Small Bets course from Daniel Vassallo, and he shared this idea of starting with low expectations. When you're expecting to make $100 from a product, what you do will be very different than if you're trying to make $100K. Your investment will be very different and your standard will be, well, much lower. Lowering my standards and being ok with a buggy, messy, imperfect product helped.

I made the product purely as a proof of concept and for my own use, but I was using the shipfast template so setting up a payment was a matter of adding a simple product ID, so I did.

One thing I did differently which changed the trajectory: I built the first version in a weekend and shared it on the TypingMind discord. The product was super niche, long term memory for TypingMind. Also because my expectations were super low, I priced it at $12.5, PER YEAR. My costs were basically nothing, and I didn't want price to work against me. People started buying it though, and it gave me a rush whenever anyone bought it. At the start, a sale a week would be a dream come true for me.

I started hanging around the TypingMind discord, looking for people frustrated with AI lacking memory, and I'd pitch my product whenever I saw someone like that, and each time that happened I'd end up making a couple sales.

I started talking to potential users almost immediately, whereas in earlier products, I always waited for it to be "ready", which would be a moving target.

I was also thinking much more strategically this time around. I saw that I was in a niche within a niche, typingmind users who cared enough about AI memory to pay for it. I knew this wouldn't scale, and I shouldn't put all my eggs in one basket. So I added support for ChatGPT via a Custom GPT, and started marketing my product on Reddit which is still my biggest marketing channel to date, although I'm started to scale beyond it now.

This was a home run because shortly after TypingMind guys shut down their discord channel, so if I hadn't moved before then I'd be dead in the water.

I found this super nifty little tool, Notikey, and I set up alerts for anytime anyone mentioned AI forgetfulness in my target subreddits. Reddit marketing is very challenging, and Reddit hates being marketed to, and that's what I love about Reddit. You need to be genuinely helpful to people, not just there to pitch your product. Most other marketing is "sneaky" or can be hacked through, and ends up attracting people who are genuinely not really interested in much else than making a quick buck, who will outspend you and sleaze their way through marketing their product. This doesn't work on Reddit, you'll get downvoted to hell and banned from communities. I actually reached out to subreddit mods and I asked them if they're fine with me pitching my product there, and I read this book, Simple Marketing for Smart People, they talked about marketing as education and I followed that approach. I would try to be genuinely helpful whether or not someone ends up using my product, my marketing message should provide genuine value for them. I would many times end up spending half an hour writing a comment, only to delete my product mention from it because it felt forced. This got me genuine credibility in the communities, and my organic comments ended up ranking on Google, so Reddit became an SEO engine for me without me ever planning for it or even knowing that was possible.

I also started expanding beyond ChatGPT because I had moved to Claude, then I saw people using other AI tools asking for memory, and I saw an opportunity. This was before MCP mind you, adding support for each platform was an involved process, I made a browser extension (I'm a browser extension developer at my full time day job), and by now I've added support for like 12 platforms to the extension, and hundreds via the MCP server.

There was a time when ChatGPT had memory but only for paying users, and free users wanted memory, and I capitalised on that. I also was hanging around in these reddit communities all the time because it was genuinely interesting to me, and I would have a pulse on what frustrated people with existing memory features. I added a feature called Buckets (folders) which became a USP for me, based on things people complained about on reddit.

One of the key things I'd say has been doing things that don't scale. I offer live chat support and handle support tickets manually. In a time where indie hackers are increasingly against offering any kind of customer support, I moved from email support to live chat support, lowering friction, increasing support volume 10x, but if you aren't going to talk to your customers you shouldn't be in business anyway. You're not just sitting there to take their money. Talking to users has constantly given me ideas to improve onboarding, functionality, and these improvements lead to more sales directly and indirectly. I also went out and looked for and pitched my product to individual users on Reddit, one user at a time.

This post is a bit messy, but I'm happy to share insights on any questions anyone here has. Marketing isn't easy but it can certainly be done, even by people who don't consider themselves natural marketers. You have to be ok with intense discomfort though. It gets easier over time, and the validation of seeing sales happen will eventually make it into a rewarding activity. A dollar I earn from my own products, on my own terms, is worth to me $10 I earn by working for someone else's business.