> For the complete documentation index, see [llms.txt](https://docs.1kosmos.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.1kosmos.com/identity-verification/core-concepts/ai-analysis-summary.md). # AI Analysis Summary *** ### Overview When an identity verification session fails, the question that matters most is simply *why*. Until now, 1Kosmos surfaced the raw check results. These are useful for engineers, but not always actionable for the operations teams, fraud analysts, and help desk staff who need to make fast decisions. The AI Analysis Summary adds an AI-powered analysis layer to the verification results view. It takes the underlying verification signals and translates them into a clear, plain-language summary of what happened, why the session was flagged, and what to do next. ### The verification results view For a completed verification session that has been flagged for manual review, a **status badge** at the outcome. A flagged session is one that did not outright fail but also did not fully clear; it is routed for a human decision before any approval is made. Core components * **AI analysis** — Contains everything a reviewer needs to make a decision. * **Documents** — Preserves full access to the underlying data. #### AI-powered IDV session analysis 1Kosmos generates an AI-powered summary for each Identity Verification (IDV) session, giving reviewers a plain-language breakdown of the verification outcome without requiring them to comb through raw session data. The summary is organized into three sections: **Session Overview** A concise summary of the verification session, covering the applicant, the document presented, and the overall outcome. This gives the reviewer an immediate snapshot before diving into the detail. **Failure Reasons** A plain-language explanation of any verification failures or flagged checks, identifying the cause of an unsuccessful or flagged session without requiring the reviewer to interpret raw signal data. **Key Findings & Recommendations** The likely cause of the outcome, along with actionable next steps. This is the section that drives the reviewer's decision - whether to approve, reject, or escalate for further investigation. ### Flagged checks The flagged checks section shows exactly what the AI evaluated and what it could not complete. In the example above, the image quality failure was the root cause, and because of it the following checks were not performed: * Field-level matching * Format checks * Bar code anomaly detection * MRZ parsing * Visual security feature validation The AI surfaces this as a structured breakdown, so the reviewer is not guessing at what is missing. They know precisely which checks need human eyes and which signals they can rely on. ### Decision panel The decision panel is located on the right side of the screen. Once the reviewer has assessed the AI summary and the flagged checks, they can optionally add a comment documenting their rationale, and then mark the session as **Approve** or **Reject** with a single click. --- # Agent Instructions This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com. ## Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question. Perform an HTTP GET request on the current page URL with the `ask` query parameter, and the optional `goal` query parameter: ``` GET https://docs.1kosmos.com/identity-verification/core-concepts/ai-analysis-summary.md?ask=&goal= ``` `ask` is the immediate question: it should be specific, self-contained, and written in natural language. `goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.