Centauri AI

Centauri AI extracts, enriches and structures financial documents such as credit agreements, pitchbooks and notices, delivering cited outputs, dashboards and audit trails for investment, risk and operations teams.

Centauri AI is a platform for financial document intelligence. It reads complex source material such as pitchbooks, credit agreements, term sheets, offering memoranda, statements and notices, extracts the figures and facts inside them, and produces structured output that stays linked to the line it came from. The company positions it as a tool rather than a service, aimed at investment, risk and operations teams that spend their weeks assembling numbers instead of analysing them. The product is organised into four layers that can be used separately or as a single stack. Extraction and indexing reads documents in full context, including tables, charts and varied file formats. Enrichment adds domain context such as entities, dates and business meaning. An agentic workflow layer applies the team's rules, reasons across documents and consolidates results. Delivery pushes the output out as SQL, CSV or JSON, or into live dashboards, with every value traceable to source. A side product, coSlash, gives visibility into coding agents, and a separate orchestration line addresses running many agents at once. In practice a deal team drops a package of documents into the system, each document routes to an agent suited to its type, and the extracted fields are cross-validated against one another. The output is a memo or data set with citations, confidence flags on uncertain values and an audit trail, which a human then reviews. Operations teams use the same machinery for reconciliation, trade matching and capital calls, with exceptions pushed to a person along with the evidence. Centauri AI sits in the enterprise segment rather than the self-serve one. Deployment is offered as hosted SaaS, single-tenant, or inside a customer's own AWS or Azure VPC, and the company cites SOC 2 Type II certification. Teams comparing it with generic document AI tools will notice the finance-specific framing and the emphasis on citations and human review; buyers wanting transparent public pricing or a quick individual trial will not find either on the homepage.