With this architecture, agentic reasoning is used for intake and context gathering, but every autonomous risk decision is still made by an auditable model, governed under an approved policy.
Blake WilliamsProduct Manager · Practitioner, 
Ramp
Post: Ramp engineering blog, Ramp engineering blog, June 30, 2026.
Blake Williams, Product Manager. Part of the team building Ramp's risk agents; employees describing their employer's own system, so the framing is promotional.
Checked against the source by a first reader, then confirmed independently by a second reader who fetched the source without seeing the first reader's notes (Independent review agent (Claude Sonnet), blind second pass from passage text and citation only, no repository access, 2026-10-31).
The page at https://builders.ramp.com/post/agentic-risk-operations is a client-rendered shell with no body text. The sentence is in the site's JS bundle as the article body, in the 'Agentic Risk Operations' article dated 2026-06-30. It sits just before the 'Isolated evaluations' heading and matches exactly. The post is co-authored by Blake Williams, Mueed Rehman, Vishal Sundaram and Brad Hoeweler, so it is the team's own words and Blake Williams is a listed author.
Blake Williams (Ramp), “Ramp engineering blog”, Ramp engineering blog, 2026-06-30. https://builders.ramp.com/post/agentic-risk-operations Archived at https://the-ai-record.com/passages/ramp-architecture-agentic-reasoning-used-intake-context-gathering-2026
Something wrong? Report a correction. JSON: /api/v1/passages/ramp-architecture-agentic-reasoning-used-intake-context-gathering-2026.json