LeadPuppet — I built a lead-gen tool because the existing ones were overpriced for what they actually do
I built LeadPuppet because finding leads for my own SaaS was harder than it should've been.
I tried a few of the "AI-powered" lead-gen tools out there and kept running into the same thing: you're paying for a bundle you don't need — discovery, AI personalization, AND warmed-up sending infrastructure, all locked together at a steep price. I just wanted the list.
So I built LeadPuppet: paste a URL, it analyzes what your business does with AI...
...then searches Google Places + scrapes company websites to find real, verified contacts.
No login, no lock-in — you get a filterable table and a CSV export, and you send from whatever platform you already use, paying only for what that costs you daily instead of bundling it all into one expensive tool.
Some real numbers from a recent local test: one search with a wider query budget pulled 500+ potential leads in about 30 seconds, and after adding a second extraction layer (a regex fallback alongside the AI one), verified email matches jumped from 53 to 148 — nearly 3x, for the same input.
The public version above is capped at 10 searches/day (3 queries each) since it runs against paid APIs and I'm keeping it free while I validate — not a technical limit, just a cost one. Happy to bump it up if there's real interest.
leadpuppet.vercel.app if you want to try it.
Would this actually be useful to you, or am I solving a problem only I have? And if you'd pay for something like this, what would feel fair — per search, a flat monthly fee, something else?
I’ve faced this too, and the issue is often data quality and feedback loops rather than AI itself—define a clear ICP, verify fresh lead data, track replies and meetings, use outcomes to refine targeting, and keep opt-outs and consent in mind.
Appreciate it, opt-outs are already built into every send, and ICP targeting happens through the /analyze step (it reads your site and suggests categories/cities before you search). The feedback loop part is exactly what I'm setting up now with the confidence-level tracking mentioned above.
The jump from 53 to 148 verified matches is notable, but have you measured whether those additional contacts produce comparable reply or meeting rates to the original set?
Good question, and no, I haven't segmented by confidence level yet, this batch only started sending few days ago. But that's actually the right way to check if the regex layer is pulling its weight: I'll track reply rate separately for "high confidence" (LLM-found, name+role attached) vs "medium" (regex-found, email only) once I have enough sends. If medium performs meaningfully worse, that tells me the LLM layer is doing more than just finding emails, it's finding the right kind of contact.
Nice progress. What is the next thing you are focusing on?
Right now: sending to a batch of real leads I generated with it (~20/day,
warming up a new domain) to get actual bounce-rate data — someone here
pointed out that's what matters more than raw volume, and I agree.
Beyond that, still figuring out the shape of it: keep it as a free
tool that feeds my own outreach, or turn it into something sellable —
and if the latter, whether the pitch is "cheap alternative to Apollo"
or something closer to what got suggested above (freshness/rebuild-on-
demand instead of a static list). Haven't committed to either yet,
which is honestly part of why I'm posting here.
Unbundling is the right instinct, and the piece you kept is the one with the least defensibility, which is worth planning around. Discovery from Places plus site scraping is replicable, and a list decays fast: the number that matters to a buyer is not five hundred in thirty seconds, it is what share of those addresses still accept mail in two months. If you can publish a measured bounce rate, that becomes the thing people compare you on, and it is a number the bundled tools avoid quoting. It also argues for selling freshness, a list rebuilt on demand, rather than raw volume.
Fair, and I don't disagree with the freshness point, I'll publish the
bounce rate once I have it.
But zooming out, the bigger gap for most indie hackers isn't "can I
build a scraper," it's "I have zero customers and no idea where to
even start looking for the first 10." That's the actual problem
LeadPuppet solves, not out-competing Apollo on data quality, but
giving someone with a product and no audience a starting point that
doesn't cost $50/mo before they've made a dollar.
Freshness matters more once you're doing this at volume. Day one,
just having a list at all is the unlock.