A batch API for images and documents. Send thousands of files and one prompt to a single endpoint. Klust-1 processes the whole batch and returns structured results.
[ 01 / The problem ]
01
Ten files are manageable. A thousand costs you a week. Someone opens each one and types notes by hand, and by file 200 they start missing things.
02
Prices, clauses, totals, serial numbers. All of it sits inside photos, scans, and contracts, where no spreadsheet or search can reach it.
03
Chat assistants take one file at a time: upload, ask, wait, repeat. You retype the same question hundreds of times, and there is still no batch mode.
[ 02 / The solution ]
POST one prompt and a list of image URLs, or upload a JSONL for per-file prompts.
Klust-1 runs the whole batch in parallel, async, at any scale. Poll the status.
Download one results file — a structured row per image, ready to load.
[ 03 / Use cases ]

Documents
Pull totals, dates, parties, and clauses from every scan.

Names, SKUs, and prices from every shelf shot.

Flag defects and read labels across the batch.
Screenshots, receipts, forms, IDs, charts. Klust-1 pulls the text and the fields out of any of them, as clean markdown or tables.
1,000+
Files per prompt
4×
Processed in parallel
~2s
Per file, on average
0
Images stored
[ Scale ]
Product catalogs, inspection photos, invoices, contracts, receipts, screenshots. Point the batch endpoint at your files, send one prompt, and get a structured row for every file. Klust-1, our in-house vision-language model, does the reading.
Klust-1 gets your prompt word for word, with no rules bolted on. It reads your images and stores none of them.
[ 04 / Pricing ]
Pro
Soon
For pipelines that never stop.
Enterprise
Custom
For sensitive, high-volume workloads.