Frequently asked questions

Prowlo is an MCP-native social data layer for AI agents. We crawl Reddit and X through our own residential proxy stack, embed every record for semantic search, and serve the result via MCP, REST, or webhook. Your agent gets typed JSON; your Reddit/X bill stays at zero.

Every account starts with a 14-day free trial — no card required. You get your own Watchers on the keywords you care about, served to your agent over MCP. After 14 days it's $19/mo to keep them running; if you don't upgrade, the Watchers pause and you're never charged.

No. Prowlo crawls Reddit through residential proxies, separate from Reddit's commercial API quota. You don't sign up for a developer key, you don't pay Reddit's $0.24 per 1,000 calls, and you can't get rate-limited by the official API.

One MCP install command. Your agent gets tools like search_dataset, social_search, list_watchers, and alert_create — all returning typed JSON with cursor pagination. The free trial includes MCP access.

Prowlo covers Reddit, X (Twitter), Hacker News, Mastodon, and RSS. They all share the same record schema, so your agent learns one shape. Bluesky is on the roadmap.

Apify and Bright Data ship raw scraped JSON priced by compute units or bandwidth. Syften has filter-count caps and added MCP last month as a Pro-tier feature. Prowlo ships structured records with embeddings, one simple price ($19/mo after a free trial), and MCP as the default surface — not a paid add-on.

GummySearch shut down in November 2025 because Reddit's commercial API made the math impossible for indie SaaS. Prowlo never depended on the official API — our crawl runs through residential proxies. The pricing-shock failure mode that killed GummySearch can't happen to us in the same way.

No. Prowlo is fully hosted — we run the residential-proxy crawl, anti-bot handling, embeddings, and the MCP endpoint. You generate an API key, drop one config block into Claude or Cursor, and your agent is reading Reddit. Nothing to deploy, scale, or maintain.