Open-source tracker for how AI models describe and recommend your brand
Emerged finds new products worth a look, daily: what’s new today
If you already pay an LLM to rank retrieval results, DNA turns those judgements into signal instead of throwing them away.
DNA stores one embedding per item under a key prefix, answers searches over that prefix, and takes a score from -1 to 1 on each result you get back. Embeddings come from whatever model you already use; a grid holds up to 1,024 nodes. A key is self-serve with no account: create a grid, save the key it returns once, and the quick start runs from Python or plain HTTP. Rate limits are 5 grids an hour and 20 a day, with larger capacity by email. Hashgrid reports needing 7x fewer LLM judgements than scoring a raw embedding list, with the research still to be published.
- Free
More like this
Describe the thing you half-remember and it opens the right page
Indexes your Markdown vault so a coding agent can cite it
Self-host App Store and Google Play keyword tracking, every feature included
Ask a GPU a hundred questions about one document and pay to read it once
Graph database where each AI agent writes to its own branch