AI Agents in PushMetrics are intelligent assistants that can interact with your data, run queries, create charts, send emails and Slack messages, and automate workflows — all through a conversational chat interface.

Unlike traditional automation that requires pre-configured steps, agents understand natural language requests and use the right tools autonomously to accomplish tasks.


What Can Agents Do?

Query Databases
Ask questions in plain English — the agent writes and runs SQL for you.
Create Charts
Generate interactive Plotly visualizations from query results.
Send Emails
Draft and send emails with data, charts, and attachments.
Slack Messages
Post updates, alerts, and reports to Slack channels.
Tableau Exports
Search workbooks, apply filters, and export views.
Skills & Memory
Follow saved procedures and remember workflows across sessions.

How It Works

1. YOU ASK "Show me sales by region" 2. AGENT PLANS Selects tools & writes SQL 3. TOOLS RUN SQL executes, chart renders 4. RESULTS Interactive table & chart appear

Agents have access to all the databases, integrations, and recipients configured in your workspace. Workspace admins control which tools and permissions each agent has.

Everything streams in real time — you see SQL executing, charts rendering, and messages sending as the agent works.


Getting Started

🚀
Every new workspace comes with a PushMetrics Assistant — a default agent pre-configured with common tools and permissions. Start chatting immediately from the Chat page.

To customize agents or create new ones, see Configuring Agents.


Key Concepts

Agent — An AI assistant with a name, system prompt, and set of tools/permissions.
Chat Session — A conversation thread (also called a "ticket") between a user and an agent.
Tools — Capabilities the agent can invoke — SQL queries, email, Slack, charts, etc.
Skills — Reusable procedures that agents can discover and follow.
Memory & Knowledge — Persistent files agents retain across conversations plus workspace-wide reference docs.