Uncensored AI for discovery and document review
Review the hostile documents, not a refusal about them
The documents that decide a case are usually the ugly ones. A review model that skips them is not reviewing the set.
Why the mainstream APIs fail here
Harassment claims, workplace investigations and criminal matters produce sets full of abuse, threats and explicit material. A hosted model will summarise the invoices and decline the messages the case is actually about, so the expensive human pass gets no help exactly where the volume is worst.
How Wild West API is used for this
Run the first pass through Wild West API: classification, summary and timeline extraction across the whole set rather than the polite part of it. A key per matter carries its own cap, so a runaway batch cannot spend past the number written on the key.
A working loop
- One key per matter, capped at the review budget.
- Batch the set through classification and summary without carving out the hostile documents.
- Hand the flagged subset to human reviewers with the reasoning attached.
Details are on the privacy page. Per-model prices are on /models and the pricing model is on /pricing.