GummySearch built a beloved product for 140,000 people on land it didn't own.

The pitch, name stripped

What actually happened

GummySearch was, by most accounts including its competitors', the best audience-research tool ever built on Reddit. You gave it a topic and it found the communities your buyers were in, then let you search, cluster, and track what they were complaining about and asking for. Over four years it grew from a solo indie project into something more than 140,000 founders, marketers, and investors relied on. This is not the story of a bad product. That is what makes it worth reading.

The ground it was built on had already shifted once, publicly. In 2023 Reddit repriced its data API, and the bill for third-party apps went from roughly nothing to millions. Apollo, the most loved Reddit client in existence, did the arithmetic on its new $20 million a year cost and shut down that June. The message was not subtle: access to the conversations was Reddit's asset, not the ecosystem's, and Reddit would set the terms whenever it chose.

For a while, tools like GummySearch continued inside the rules as they stood. Then the rules moved again. Reddit's data API terms restrict commercial use, and in late 2025 that restriction landed where it was always eventually going to land. The founder's shutdown note is unusually honest about the cause: the product was closing 'to comply with Reddit's API policies, which forbid commercial applications.' No pivot, no acquisition, no slow fade. A working product with six figures of users, switched off by a policy it had no vote in.

The shutdown itself was handled with real class: paying subscribers kept access through their billing period, lifetime-deal holders kept it for a further year, and the founder published an alternatives page pointing his own orphaned users at other tools. But look down that list and the uncomfortable pattern repeats. Nearly every alternative on it is built on the same foundation: the platform's API, or automation acting on the platform's surface, each one a policy decision away from the same ending.

That is the part that outlives GummySearch. The category keeps producing tools that treat a platform's conversations as raw material they are entitled to process, and the platform keeps demonstrating, in 2023 and again in 2025, that it disagrees. Every one of these products carries the same unpriced liability on its balance sheet, and most of their landing pages never mention it.

The lesson

We fed the stripped pitch to Kasspian's scorer with no product name attached. It scored it 5/10, said test it first, and named the killer without being told the history: 'if the platform throttles, paywalls, or kills the API, your entire product evaporates overnight with no recourse.' The model had never heard of this particular product. It didn't need to. The risk was visible in the pitch's own architecture, which means it was visible on day one.

Platform risk is not a footnote risk, because it is uncapped. A pricing mistake costs you margin; a churn problem costs you growth; a platform policy change costs you the company, all of it, on a date you don't choose. The honest question for any business built on someone else's data or distribution is not 'is this allowed today' but 'what happens to me on the day it isn't', and if the answer is 'everything ends', that is not a risk you are carrying, it is a decision you have deferred.

The test, as ever, is cheap and nobody runs it: read the platform's terms as if you were their lawyer looking for a reason to shut you down, and price what compliance-by-architecture would cost you instead. Some products in this category can exist without touching the API at all: reading public pages, reasoning about communities instead of bulk-processing them, keeping a human's hands on anything that posts. That architecture is slower and humbler, and it is the one still standing. Full disclosure: Kasspian competes in this category, and made exactly that trade, which is why this teardown exists and why the alternatives page it would love to be listed on is full of products that haven't.

Common questions