All projects
02

Getrive

B2B SaaS that watches Reddit and Hacker News for people describing the exact problem your product solves, scores each post with AI, and drafts an honest reply for you to review before it ever goes out. Live at getrive.app.

Next.jsPrismaStripeAI SDK
Getrive landing page showing a flagged Reddit post next to its AI-drafted reply

Overview

Cold outreach and generic automation burn trust fast, and most early-stage founders genuinely don't know where the people who need their product are already talking about the problem out loud — so they either spam strangers or say nothing.

Getrive runs a continuous ingestion scheduler across Reddit, Hacker News and IndieHackers, scores every post against the product's stored positioning with AI, and drafts a contextual reply — but nothing ever sends itself. Every draft sits in a review queue until a human approves it, and every approved send is attributed back to signup.

Features

  • Continuous, multi-source monitoring across Reddit, Hacker News and IndieHackers
  • AI relevance scoring of every post against the product's stored positioning, not keyword rules
  • Contextual, on-brand draft replies generated per matching post
  • Human-in-the-loop review queue — approve, edit or skip; nothing sends automatically
  • Signup attribution: every approved reply carries a tracked link back to conversion
  • Early-access pricing locked in for first users, billed through Stripe
  • A published guides section with the underlying "finding your first users" playbook

Technologies

  • Next.js 16 (App Router), React 19
  • Prisma 7 + Postgres — typed data layer
  • NextAuth v5 — authentication
  • AI SDK (Anthropic + OpenAI) — relevance scoring and reply drafting
  • Stripe — subscription billing (currently free during early access)
  • Resend — transactional email
  • PostHog — product analytics
  • Sentry — error tracking
  • Playwright — end-to-end test coverage

Technical highlights

  • Fair, continuous ingestion scheduler polling multiple sources oldest-checked-first, rather than periodic batch scraping that would starve slower sources
  • Positioning-aware AI scoring, not keyword matching — the system reasons about whether a post genuinely describes the product's problem, not whether it contains certain words
  • Hard architectural separation between draft generation and send — there is no code path for autonomous posting, by design ('auto-reply systems disabled by design' is an actual constraint, not just copy)
  • Per-reply tracked attribution links (/r/{id}) connecting a specific Reddit/HN comment to a specific signup
  • GDPR-oriented documentation (a RoPA and a DPIA specifically covering the signal-monitoring pipeline) written because the product processes public posts at scale

Status

Live in early access at getrive.app, free during early access with pricing locked in for early users. 67 commits, currently iterating on usage-based pricing on a feature branch.