A no-code platform for turning approved knowledge—FAQs, documents, links, and facts—into a focused, shareable AI agent. You define the agent\'s purpose, audience, allowed and blocked topics, tone, uncertainty response, and when to hand off to a human; then publish it at qlynk.site/username or embed it on an existing website with a script . Knowledge sources include PDF, DOCX, TXT up to 3 MB, plus manual facts and public links . The dashboard records conversations, analytics, and knowledge gaps so owners can improve answers over time . It is MIT-licensed and self-hostable, with hosted Creator free forever (one agent, 5,000 messages/month, no card) and Agency at $38/month or $360/year for 10,000 messages/month and removal of the \"Powered by Qlynk\" label . Critical caveats: Independent third-party review volume is essentially absent—no G2/Capterra/Trustpilot data surfaced. The product is very young (Product Hunt launch and #1 on Smol Hunt in August 2026) . The vendor\'s own blog candidly states that generative AI can still hallucinate and that \"no general-purpose language-model system should promise\" zero hallucinations . And the hosted plans cap files at 3 MB with no separate storage allowance for Agency .
Pros & Cons:
Pros: The approved-knowledge-first design is a concrete differentiator—every agent is scoped to a defined job with explicit blocked topics, an uncertainty response, and human handoff rules, rather than being an unrestricted general assistant . The free Creator plan is genuinely usable: one agent, 5,000 messages/month, no expiry, no card required, and the old trial accounts were migrated to it without losing knowledge or conversations . The embedding flow is well-documented and security-conscious: each widget installation uses an opaque identifier, restricts loading to exact allowed origins, and changing the public username does not break the embed . The product is MIT-licensed and self-hostable, with setup documentation and the source on GitHub . The dashboard\'s knowledge-gap tracking and conversation review give owners a practical improvement loop . And the vendor\'s own writing on RAG and hallucination is unusually candid about limitations—it explicitly says approved knowledge improves grounding but does not guarantee factual answers .
Cons: Independent validation is essentially absent—no substantive third-party reviews surfaced, and the product is very young (launched August 2026) . The hosted file limit is 3 MB per document with no additional storage on Agency, which constrains knowledge bases built from large PDFs . The free plan\'s 5,000 messages/month pause at the limit rather than offering overage, and upgrading does not reset already-used messages for the month . The public URL is not permanent—changing the username stops the old address immediately, with no redirect and a 30-day cooldown . And while the self-hosted option exists, it requires your own infrastructure and provider credentials, with all costs borne by you .
Who is it for?:Freelancers, experts, small businesses, property/venue operators, and support teams that repeatedly answer the same approved questions and want a focused, branded AI agent without coding or a card. Users who want to embed an agent on an existing website or share a hosted link, with explicit boundaries and human handoff. Developers or teams that prefer self-hosting an MIT-licensed agent over a closed SaaS. Not suited for anyone requiring extensive independent third-party reviews before adoption, teams that need to ingest large documents beyond 3 MB, users who want unlimited messages on a free plan, or organizations that need a permanent public URL without a 30-day cooldown on changes.