# Research Agent The ClearFeed Research Agent is an AI Agent that helps you explore and analyze the data stored in your ClearFeed account using natural language. ## What the Research Agent Does Research Agent can help you: * Explore requests in natural language * Investigate trends across queues, channels, teams, and customers * Answer questions similar to what you would normally check in the Insights dashboard * Find relevant requests by filters or search terms * Summarize what is happening in your ClearFeed account based on current request data Typical use cases include: * Reviewing request volume and resolution trends * Investigating pending or high-priority work * Looking for customer-specific patterns * Finding requests tied to a topic, source, or integration and understanding overall trends * Spot-checking service levels like first response time, resolution time, or SLA breaches ## What Data It Can Query Research Agent currently works with the data available inside your ClearFeed instance. It does not browse the web or answer from your knowledge base. Today, it can query data such as: * Requests and their metadata * Status, timestamps, channel, collection, requester details, and message history when needed * Request search results * Including keyword and semantic search over request content * Collections and channels * To understand queues and request sources * Teams * Including team membership context * Customers * Including customer records and customer-linked request analysis * Custom fields * For request and customer attributes configured in your account * Ticket forms * Where relevant to request workflows * Insights and reporting metrics * Request counts, solved counts, response times, resolution times, SLA metrics, CSAT metrics, and related operational analytics ## How It Works Under the hood, the ClearFeed research agent is built using an agent harness that has access to several tools. Tool calling means the model does not answer only from its own reasoning. Instead, it can decide when it needs to access structured data in ClearFeed, pass structured arguments to that tool, read the returned data, and then use that data to construct a grounded response. This allows the agent to translate a natural language question into one or more API-backed queries against your ClearFeed account. For example, one of the tools that is essential to the agent is the search tool which can search over your ClearFeed requests. A query like 'Find me Zendesk related issues in February would map to a search query like this ``` { "query" : "Zendesk issues", "created_after" : "2026-02-01", "created_before" : "2026-02-28" } ``` In practice, any functionality that is exposed through the supported ClearFeed APIs can also be performed through the Research Agent in natural language. Rather than manually calling the API or navigating filters in the web app, you can ask the agent what you want to know and it will decide which tools to use to retrieve the relevant data. The agent currently has access to the following tools: * Collections lookup * View collections and linked channels * Requests listing * Filter requests by time, status, collection, channel, requester, and more * Request lookup * Open a specific request and inspect its details * Request search * Search requests using keyword and semantic search * Teams lookup * List teams and inspect team membership * Customers lookup * List, search, and fetch customer records * Custom fields lookup * Inspect request and customer custom fields configured in ClearFeed * Ticket form lookup * Read ticket form definitions * Insights query * Run analytics queries across request and service metrics The agent only has the ability to perform read actions. It can investigate and report on issues, but it does not update requests, post replies, modify ticket fields, or take actions in external systems. ## How to Create a Research Agent {% hint style="info" %} The Research Agent feature is currently in beta. {% endhint %} To create a new agent of the Research type, go to: ![Research Agent creation screen](https://3455705434-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FE2O2wTaNovd6fXpEuLKz%2Fuploads%2Fgit-blob-930a0cb76a2614e4b3a3e137c92ce33cae43f250%2Fresearch-agent-create.png?alt=media) When creating the agent, you can configure a base user prompt. This lets you add account-specific context and instructions, such as how your team names queues, what business context matters most, or any guidance you want the agent to keep in mind while answering questions about your ClearFeed data. ## Example Questions Here are examples of queries that should work well with Research Agent: * Show me all high-priority requests created in the past week that are still pending on us. * How many requests were solved in the last 30 days? * Which customers created the most requests this month? * What was our median first response time last week? * Find requests mentioning SSO failures or login issues from the past month. What are the broad steps we are suggesting to users to resolve these queries? * Categorize the requests from the last month and tell me the major themes. Make sure to provide a few example request IDs for each category. The Research Agent helps teams chat with their ClearFeed data. It is meant to be used for operational questions, analytics, request discovery, and account-level support insights.