I solo-built Heardley, a Reddit lead generation and monitoring tool for founders and marketers. It scores threads for buyer intent, drafts replies in the user's voice, and scrubs AI tells before posting. $9 a month or $54 lifetime, bring-your-own-key for free AI. I built the product, the platform, the brand, and the marketing alone.
Heardley is a Reddit lead generation and monitoring tool. The pitch in one line: 500 Reddit threads in, 5 worth a reply out. The product watches every subreddit a founder cares about, scores each thread against a buyer-intent rubric, and surfaces only the threads where a human reply has a real chance of converting.
The rest of the workflow is built around what happens after a thread surfaces. Heardley drafts a reply in the user's own voice using a tuned prompt and prior writing samples, then runs the draft through an AI-tell detector that scrubs the patterns that get Reddit posts downvoted and reported. The user reviews, edits, and posts. The point is to take Reddit lead-gen from an unwinnable time sink to a 15-minute-a-day habit.
Heardley is solo-founded. I built the entire product, end to end. The Reddit ingestion pipeline, the buyer-intent scoring model, the voice-tuning system, the AI-tell scrubber, the dashboard, the billing integration, the marketing site, the SEO architecture, the schema markup, the topic clusters, the email lifecycle, and the support flow are all my work.
There is no co-founder, no team, no investor pressure. The product is small on purpose. Pricing is $9 a month or $54 lifetime, and users bring their own AI key so the unit economics work at any scale without me touching the price.
AI vibe coding is how I shipped it fast. Five years of LLM production workflow experience compounds when you are building a product like this: Claude and Cursor running constantly, custom prompt libraries for routine work, drafts AI-assisted and human-reviewed against documented rules.
Heardley is a server-side SaaS. A scheduled worker polls subreddits the user is tracking, scores each new thread with a small intent-classification prompt, and queues high-scoring threads to the dashboard. The user reviews queued threads, opens the AI-drafted reply, edits it, and posts to Reddit through the standard Reddit web interface.
Most Reddit tools either alert on too much (every keyword match floods the inbox) or alert on too little (sentiment filters miss buyer-intent threads). Both fail the same way: the user spends more time triaging the tool than posting. Heardley's whole design is one filter early in the pipeline, doing the hard work, so the user's inbox is small enough to actually clear.
Building solo lets that call stay consistent. Every feature is judged against one question: does this make the user's daily Reddit time shorter or longer? Features that lengthen the workflow are cut, even when they would obviously sell.
AI-powered 2D game studio SaaS. I built the early prototype with Tony Valcarcel before the full founding team came together.
Local-first AI video studio for faceless YouTube creators. Built with Cesar Martinez and Zac Dillard.
Flat-fee websites for local service businesses. Same team as Phantomline: me, Cesar, Zac.
Four years scaling community to 170k+ followers. Contributed to the largest NFT mint in OpenSea history at 220k assets.