Understand Agent capabilities

The Agent is a tool-using underwriting partner, not a separate chat box. It can work across the deal workspace you are already using, then leave a visible trail of actions, sources, and follow-up context in the transcript. ## What the Agent can use In normal underwriting work, Reia can search processed files, read relevant passages, tables, and images, inspect Inputs and Statements, trace Graph relationships, run sensitivities or rollups, update editable assumptions when you ask for that, and draft or edit working documents when you ask for a persistent document change. Some chat setups can also use web research, code-assisted analysis, or image generation. When those capabilities run, the transcript shows an action row such as web search, code execution, or image generation so the work is easier to follow. ## How to direct Reia Describe the job in underwriting language. Name the goal, the source material or panel to use, the expected shape of the answer, and whether you want sources, a memo-ready section, a table, a list of diligence questions, or an input update. You do not need to choose tools by name. A request like "Compare the rent roll to the T-12, cite the support, and list the assumptions that should change" gives Reia enough direction to search evidence, inspect assumptions, and produce a reviewable next step. ## Work iteratively Agent work compounds when you keep the thread focused. Start with one question, review the action and source trail, then ask a narrower follow-up: show the source for a number, trace the metric to its inputs, update one assumption, or draft the conclusion into a working document.