Hey Product Hunt. I'm Mariale, co-founder and CTO of KINETK. We're opening our core Graph Service API to the PH community today, and I want your feedback. The backstory. The internet is no longer text-first. It moves through clips, images, screenshots, and visual memes. Almost all data infrastructure still treats text as the universal medium. It indexes surface-level captions and tags. The moment you bolt multimedia onto a text-first index, you inherit real architectural debt. Clumsy cross-system lookups. Split database paths. Dedup pipelines that never realize a video clip and its screenshot are the same trend. We made one load-bearing decision to fix that. Enforce a single shared embedding space across video, image, and text from day one. Then we did the part a vector DB cannot. We connected it as a graph. Why a graph, and not just vectors. Vectors tell you what content is about. The graph tells you how it is connected. Shared audiences. A narrative jumping from Reddit to TikTok before it hits X. Who is amplifying versus originating. Combine both in one query layer and you get retrieval that is cited and auditable instead of a probabilistic guess. Under the hood, you get direct access to the raw data layer. Not another chatbot wrapper. - 15B+ multimodal vectors. Text, image, and video frames in one space for instant cross-modal lookups. - 500M+ enriched metadata points. Platform-native, structured into clean namespaces for precise filtering. - 100M+ core records. A production-scale corpus, continuously refreshed by our global Sentinel network. Two ways in. The HTTP API, REST plus a key, for backends and pipelines. And the kinetk-mcp server, drop-in tools for Claude, Cursor, and Gemini CLI, so your agent fetches grounded context in one call. No token guesswork. Token pricing is broken for data lakes. It monetizes compute, not data quality. KINETK is credit-based and pay-as-you-go. Every response returns an X-Kinetk-Credits-Remaining header, so your agent fleets manage their own egress budgets in code. Launch offer. Sign up today and get $50 in free API credits. Go break our infrastructure. Ready for production? Get a 50% bonus on your first credit top-up. It is live and self-serve. The quickstart is at http://docs.kinetk.ai . I'll be here all day. Tell me what you would query first, and where you would want it to go next.

KINETK API

Real-time multimodal intelligence API for AI agents

Real-time multimodal intelligence API for AI agents