What made you build your product? Drop the real story below.
I'll go first.
Six months ago I was running growth for a small team. Seven tools open at once. I'd draft something in Notion, paste it into the CRM, copy the numbers from Analytics into a Sheets report nobody read, then schedule the post in a different tab. Every week the same loop. Nothing talked to each other, so I was the integration layer.
I kept thinking the problem was that I needed a better tool. It wasn't. The problem was the handoffs.
So we built Dolv. An AI operator that works across the tools you already have, not a replacement for them. It plans the work, does the work, and holds anything customer-facing in an approval queue until a human says go. Because unsupervised sending is the thing nobody actually wants, they just put up with the alternative.
Product Hunt page: https://www.producthunt.com/products/dolv
Your turn. Curious what the real origin story looks like for others:
What problem pushed you to build?
Who kept telling you it was painful before you finally listened?
What's the part you still haven't solved?
Will read every reply.
Replies
Great thread format, answering all three.
What pushed me to build: Twenty-five years running engineering organizations, and my own household still ran on the same duct tape as everyone else's: a budget spreadsheet, a calendar, a folder of insurance PDFs, bills in four apps, and all of it living in my head as the single point of failure. The moment that broke me was realizing my family couldn't answer "can we afford this trip?" without me personally assembling data from five places. I was the integration layer at home, same as you were at work.
Who kept telling me it was painful: My wife, for years, every time a question about our own life required "let me check three things and get back to you." And honestly, my own layoff told me loudest. When a private equity acquisition ended my VP role this spring, the first project I reached for was the one my household had been requesting forever. Built it solo in four months with AI agents as my engineering team.
What I still haven't solved: Onboarding. Every user who tries Kaevo wants what it does, and the first ten minutes of setup is where they stall, because a household OS is only useful once your real life is in it, and getting your real life into it is work. Universal pattern, no clean answer yet. If anyone here has cracked the cold-start problem for data-heavy personal tools, I'm listening.
Kaevo (kaevo.ai), the AI household operating system. Launching here on the 22nd. And Dolv's approval-queue insight is right, by the way: "unsupervised sending is the thing nobody actually wants" applies to household finance actions too. Good luck with the launch.
Progress AI Observability
@michael_shollenberger just passing by to say this sounds super cool and I really, really relate to the pain. I'll defintely check it out, good luck on your launch and beyond
@denitsapenchevavaltchanova Thank you! And 'I really relate to the pain' is basically our entire market research summarized in five words, ha. Would genuinely love to hear what you think after you poke around, especially anything that feels clunky in the first ten minutes. That's the part I'm obsessed with right now. Good luck with your own launches too! 🧭
The honest version: I kept running into the same question everywhere – "what colors actually suit me" – and the two available answers were a five-question quiz that never looks at you, or a $200–500 in-person session where two analysts can hand the same face different verdicts. That gap annoyed me enough to build the thing that just… looks: one selfie → your 12-season type, undertone, gold vs silver, plus a lab that grades the makeup you're actually wearing. Funny detail: the color half wasn't even in v1 – users kept sending screenshots of other tools' color results next to our makeup reads, and the effort of that workaround was the roadmap telling me what it wanted.
It's siOsi.me, free – launching here on the 22nd.
What pushed me to build: I was running SEO for a handful of small sites and spending close to an hour every morning just checking Search Console and Semrush by hand across all of them, then writing the same report structure over and over. The analysis itself was fine, the repetition was the actual problem.
Who kept telling me it was painful: mostly my own calendar. I kept blocking that hour for something that should not have needed a human doing the checking part.
What I still have not solved: attribution. I found recently that a chunk of one site's backlink growth traced back to a single rotating ad widget, not real editorial links, and separating real signal from that kind of noise is still a manual judgment call, not something I have automated yet.
hrefStack (hrefstack.com). Still working on the parts that need judgment, not just checking.
I started building Sentra because of something I kept seeing in healthcare, a lot of harm doesn’t come from a lack of effort. It happens because important things are missed at the wrong moment.
A drug interaction nobody notices. A deteriorating patient who doesn’t look critical yet. A diagnosis that comes too late. A referral that should have happened earlier.
In primary care, especially where doctors work with limited time, limited diagnostics, and a constant stream of patients, those small misses can become very serious.
That’s the problem behind Sentra, an AI system I’m building for healthcare.
The idea is not to replace doctors or make AI the decision-maker. I want to build something that works beside clinicians—a second layer of attention that can flag risks, help think through differential diagnoses, reduce administrative work, and hopefully catch things before they become preventable harm. That human-authority principle is also explicit in Sentra’s strategic mandate.
There’s still a lot I haven’t solved. Clinical validation is hard. Trust is hard. Building something that actually fits into a doctor’s workflow is much harder than building a good demo.
But the reason I keep working on it is simple:
Too many preventable outcomes begin with something small that was missed. I want to see if we can make those moments harder to miss.
I was coming up to the end of the financial year. Hadn't yet filed my taxes or done my accounts and the stress was getting to me, mountains of receipts and transactions across multiple bank accounts. I tried setting up with Xero but it was down at the time (wouldn't let me create an account). I tried using google sheets and got frustrated so I built @Wrightly for myself and now taking on feedback from real users to improve it for them too.