Parsewise set a new state of the art on Databricks OfficeQA, a grounded reasoning benchmark over 89,000 pages of dense financial documents, outperforming frontier models like GPT-5.5 and Fable 5.
Public source
Publisher name
Public post
Thrilled to share I have joined Parsewise as founding GTM lead. Parsewise has built the most robust multi-document reasoning tool I have seen in the market. Messy docume…
Company
Parsewise
The agentic decision platform to manage casualty, specialty, KYC, and financial risks better. YC X25
- Industry
- Software Development
- Location
- London, GB
- Company size
- 2–10 employees
About Parsewise
Parsewise is a decision platform for insurance, reinsurance, asset management, and compliance teams working with data-heavy portfolios. We help underwriters, claims leaders, and M&A teams surface severity drivers, reconcile loss data, and accelerate portfolio underwriting with structured, traceable AI analyst agents. The result: improved pricing accuracy, reduced claims leakage, and higher combined ratios.
See moreLatest activity
Latest activity from Parsewise
8 signals
Products & Services
Parsewise engineering team pushed the frontier of grounded reasoning technology with Navi making it accessible to any business user
Products & Services
Parsewise introduced Navi as a new interface to the Parsewise Data Engine to allow users to begin with the questions they wanted answered about the data sitting in their documents.
Products & Services
Parsewise added new features to Navi, including replaying the decision process across data pipelines, talking to Navi via speech, and sending screenshots to share diagrams.
Discover more
Similar signals
Similar public activity from other companies.
Products & Services
LlamaIndex
LlamaIndex improved its document parsing capabilities across the board in the past three months, including its LlamaParse cost-effective mode.
Products & Services
Reducto
Reducto released r-1, the best parsing model yet, achieving 20% lower error than previous agentic models, faster processing, and 1¢ per page up to 6× cheaper from earlier.
Products & Services
Multiverse Computing
Multiverse Computing introduced Quasar 438B, the top European AI model evaluated in the latest Artificial Analysis benchmark comparison.
Products & Services
Loop
Loop launched its first benchmark, AuditBench, comparing frontier LLMs with its domain-specific AI harness on real-world invoice reconciliation and audit tasks.
Products & Services
Vals AI