AI Insights

Overview

DeepTraq's Insights Assistant turns code-scanning findings into concise, actionable intelligence using AI. From a scan result (for example, the Chef demo scan), you can open the Insights Assistant next to the code to generate a structured analysis of a vulnerability without leaving the platform. The assistant can breakdown technical details, attack vectors, and affected components, surface relevant external sources, and keep an auditable chat history of your investigation.

---

Supported Platforms

• Code-scanning workflows that surface vulnerability findings (example shown: Chef demo scan).
• Model-agnostic AI engines — selectable at runtime (examples shown in the UI: ChatGPT, Claude, Gemini).
• Integration with web search tools to ground AI responses using real-time internet sources (security blogs, exploit databases, CVE references).

Prerequisites

• A completed code scan with at least one vulnerability finding (the Insights Assistant opens from a scan result).
• Insights Assistant feature enabled in your DeepTraq instance.
• Access to at least one AI model configured in the model selector (ChatGPT / Claude / Gemini or other supported engines).
• Web search tool enabled if you require the assistant to use real-time internet sources for grounding and source linking.
• Sufficient user permissions to view scan findings and open the Insights Assistant.

Steps

1. Open the code scan that contains the vulnerability findings .

2. From the scan results, click the DeepTraq AI Insights icon adjacent to the specific finding to launch the Insights Assistant.

3. (Optional) Choose a pre-built prompt template to accelerate triage — templates are provided to standardize investigation prompts.

4. Open the model selector and pick the AI engine you want to use for the analysis (examples: ChatGPT, Claude, Gemini).

5. Submit a prompt asking the AI to break down the technical details, attack vectors, and affected components for the selected issue.

6. Review the AI-generated structured summary that appears beside your code — it will include primary risks and impact.

7. If you enabled the web search tool, inspect the direct links to external sources the assistant used (security blogs, exploit databases, CVE entries) to validate the intelligence.

8. Use the chat history panel to review prior investigation sessions or continue a previous deep dive — all assistant interactions are automatically logged.

Field Reference

Field	Description	Example
DeepTraq AI Insights icon	Launches the Insights Assistant for a specific finding directly from the scan results.	Click the AI icon next to a vulnerability to open the assistant.
Pre-built prompt templates	Ready-made prompts to speed triage and standardize investigations.	"Summarize technical impact and attack vector for this finding."
Model selector	UI control to choose which AI engine to use for the analysis.	Select "ChatGPT" or "Gemini" from the dropdown.
Structured summary	The AI-generated, easy-to-read breakdown of risks, affected components, and impact.	"Primary risk: remote code execution; Affected: library X v1.2.3"
Web search tool	Optional capability that allows the AI to query the internet and cite sources for grounding.	"Use web search to find exploit write-ups and CVE pages."
Source links	Direct links to external references used by the assistant to generate its analysis.	Link to a specific CVE page or security blog post cited by the AI.
Chat history (Investigation log)	Automatically saved conversation history for each Insights session to allow review and continuation.	Past session showing question, AI response, and timestamp.