Domain-specific enterprise search infrastructure
Product manager at Cohere on enterprise AI search infrastructure and deep research agents
This points to search shifting from finding pages to traversing trusted systems of record. In regulated fields, the hard part is not ranking web links, it is pulling answers from the right corpus, like journals, filings, case law, and internal documents, then reasoning over them in a workflow that can stand up to scrutiny. That is why Cohere is pushing North toward secure enterprise grounding, while vertical tools like OpenEvidence and Hebbia are winning by going deep in one domain.
-
Cohere started as a model company, but North shows where it is headed. North is an enterprise assistant for knowledge workers, deployed in private environments, and its current priority is getting internal document grounding right for large enterprises with messy systems like SharePoint and proprietary data stores.
-
The product logic behind domain specific knowledge bases is simple. Open web search returns a lot of recycled SEO content, while specialized work needs primary materials. In medicine that means journals and textbooks. In finance it means filings and analyst reports. In law it means precedent, statutes, and case documents.
-
The market is already splitting this way. OpenEvidence built a medical search engine on clinical literature and journal partnerships. Hebbia is used after general search when finance teams need to run diligence, contract review, and memo creation over large document sets. Cohere is trying to supply the full stack underneath similar enterprise workflows.
The next step is that search vendors become data access layers for agents, with each serious workflow tied to a specific corpus and set of connectors. The winners will not be the tools with the biggest web index. They will be the ones that combine trusted domain data, secure enterprise access, and enough reasoning to turn retrieved material into work products.