Question 1

What types of documents can ClearSight process?

Accepted Answer

ClearSight processes PDFs, scanned documents (via OCR), and structured text files. It supports domain-specific documents like Scheme Information Documents, CAS statements, policy wordings, financial statements, and regulatory filings. New document types are added through YAML configuration — no code changes needed.

Question 2

How does ClearSight avoid hallucinations?

Accepted Answer

ClearSight uses a deterministic-first pipeline. Over 70% of extraction happens using rule-based methods (pdfplumber for text, camelot for tables, OCR for scans) — with zero LLM involvement. When LLMs are used for verification, every claim is cross-referenced against the source text with page-level citations. A separate verification step catches discrepancies.

Question 3

What does the free trial include?

Accepted Answer

The 15-day sandbox gives you full API access with Tier 0 extraction (deterministic, $0 cost). You get pre-loaded sample documents, a Postman collection, and OpenAPI documentation. No credit card required. Process up to 10 documents per day.

Question 4

How much does processing cost?

Accepted Answer

Tier 0 (deterministic extraction) is $0 — it handles 70%+ of processing. When LLMs are needed for verification or synthesis, costs scale based on tier: Tier 2 at $0.15/M tokens, Tier 3 at $3/$15/M tokens. Average cost per document is under $0.05. You set budget caps per tenant.

Question 5

Can I add my own document types?

Accepted Answer

Yes. ClearSight's domain repository system uses YAML configuration files to define document types, extraction rules, validation schemas, and lens configurations. Adding a new document type requires no code changes — just a new YAML definition.

Question 6

Is my data isolated from other tenants?

Accepted Answer

Yes. ClearSight uses PostgreSQL Row-Level Security (RLS) enforced at the database level on every table. Tenant isolation cannot be bypassed by application code. Each tenant's data is cryptographically separated.

Question 7

What's the integration effort?

Accepted Answer

ClearSight is API-first. A single POST to /v1/documents/upload processes a document end-to-end. You get structured JSON back with extracted data, verification scores, and citations. Most integrations are live within a day using the Postman collection.

Insights & Updates

Introducing ClearSight: Document Intelligence for Financial Services

Why Deterministic-First Beats LLM-Only for Document Extraction

Building a Zero-Hallucination Verification Pipeline