Semantic Search & RAG
Give your own documents a search that understands meaning. Embed a corpus, run semantic search and answer questions with citations back to the source — the building blocks of retrieval-augmented generation over your data. Everything through one API, priced per use.
Get text embeddings
Turn text into numbers that capture its meaning, so 'cancel my plan' sits next to 'terminate subscription' even though they share no words.
$0.002
Embed a whole corpus
Vectorize ten thousand product descriptions in one submission instead of ten thousand separate calls: you send the whole list, track one job and collect every…
$0.002
Index your text
Skip the four decisions a vector database demands, from how to split text to where to host the index: send your wiki pages, help articles or notes and get an…
$0.002
Index PDFs and scans
Find 'the clause about early termination' without knowing which contract holds it: your PDFs, scanned pages included, are read page by page and turned into an…
$0.010
Index a website
Someone asking 'how do I change my billing cycle' should find the page titled 'subscription management': give one starting address, the crawl follows the links…
$0.040
Search by meaning
Someone who types 'forgot my password' finds the article titled 'resetting your access credentials': results are ranked by closeness of meaning, not by shared…
$0.002
Hybrid search
One search box that handles both 'SKU-88213' and 'lightweight running jacket': exact terms keep their precision, loose descriptions are matched by meaning, and…
$0.002
Rerank results
Move the right answer from position eleven to position one: send a query with the candidates your search already found, and they come back reordered by how well…
$0.002
Find similar documents
Start from the contract, article or ticket you already have and get the documents closest to it in meaning: the whole document is the query, so there are no…
$0.002
Reverse image search
Upload one photo and find the images in your library that resemble it, even a duplicate shot from a slightly different angle, with no captions, tags or file…
$0.002
Recommend related content
Add a 'you might also like' row to fifty thousand items without anyone hand-picking: each item is compared by meaning with the candidates, so a product launched…
$0.002
Cluster texts
Turn ten thousand tickets, reviews or survey answers into a short list of themes: rows that say the same thing land in the same group, and you see how large…
$0.002
Detect outliers
Find the handful of rows that do not belong among ten thousand normal ones, such as a mislabeled entry, an unusual claim or a description copied from the wrong…
$0.002
Match products
'Wireless Mouse Blk 2.4G' in a supplier feed and 'Black Wireless Optical Mouse' in your catalogue are the same item: this finds such pairs without a shared code…
$0.002
Ask your documents
Ask what your benefits handbook says about leave and get an answer drawn only from that handbook, with the passages it used; if your documents are silent, it…
$0.003
Answers with citations
Ask about a benefits rule and receive the answer plus the exact clause it came from, ready to attach to a ticket, so every claim can be checked in seconds.
$0.003
Chat with a PDF
A 90-page equipment manual has the answer on page 47; ask your question and get it straight from the file, with the passage attached, instead of reading the…
$0.003
Chat with a website
Ask your store's bot about returns and it answers from your real policy page, read again each time someone asks, never from an invented, more generous policy.
$0.003
Build an FAQ bot
A customer asks about a password reset in their own words and gets the answer from your help article, even if they phrase it differently; questions your FAQ…
$0.003
Rewrite queries
A shopper types “warm jacket cheap”; this turns it into a clearer, more searchable query before your catalog search runs, so the same search finds more relevant…
$0.003