How To Use DataPeak Agentic AI
Build custom workflows or use ready-made AI agents to automate your operations.
The Agentic AI section is where Admins and Sub Admins create powerful multi-step automations (Custom Agents) or use DataPeak’s pre-built System Agents. This guide explains how to access the page, create custom agents, connect components, run workflows, and use or manage System Agents.
1. Accessing Agentic AI
Sign into DataPeak
Click Agentic AI in the left-hand sidebar
You’ll see two main sections:
Custom Agents (at the top)
System Agents (below)
Only Admins and Sub Admins can create or edit agents.
2. Custom Agents
Create your own AI-powered workflows step by step.
Custom Agents allow you to design fully automated workflows using drag-and-drop components. These can summarize data, query datasets, merge files, send outputs, create documents, and more.
2.1 Creating a Custom Agent
Go to Agentic AI
Click Create Agent
You’ll enter the Agent Builder interface
This is where you assemble your workflow.
2.2 Understanding Components
On the left side of the screen, you’ll see all available components organized from top to bottom. These include:
Input components (datasets, text, parameters)
Processing components (summaries, queries, transformations)
Output components (PDF, file downloads, responses)
All custom agents require at least one input component.
To add a component:
Drag a component onto the canvas
Repeat for all components you need
To configure a component:
Hover and click the gear icon
Or click Add Parameters on the component
Each component may have required configuration settings.
2.3 Linking Components
To create a workflow:
Connect the green output circle of one component
→ to the black input circle of another component
You can link multiple components to build branches or sequences.
2.4 Running a Custom Agent
Once your workflow is built:
Click Run (top-right)
Wait for the workflow to process
Each component will show a green checkmark if successful
Viewing results:
Click any green checkmark to see output from that stage
If your workflow includes a Download component (PDF, file, etc.)
→ You’ll be prompted to download the file immediately
2.5 Saving a Custom Agent
When your workflow works as expected:
Give your agent a name
Add a description
Click Save
Your saved agent appears under the Custom Agents section.
2.6 Using, Editing, or Publishing a Custom Agent
Click any saved custom agent to see actions:
Use/Edit Agent — opens it in the Agent Builder
Delete — permanently removes it
Publish / Unpublish — controls whether others can use it
Publishing makes the agent available to other users in the workspace.
3. System Agents
Ready-made agents built by DataPeak.
System Agents are prebuilt workflows provided by DataPeak to automate common tasks across your workspace.
They require no setup and are immediately usable.
3.1 Accessing System Agents
Go to Agentic AI
Scroll down to System Agents
Browse the available pre-built agents
Each agent includes:
A title
A short description of what it does
A Use Agent button
Categories may include:
Data cleaning
Data enrichment
Data integration
Monitoring
Resource allocation
Foundational workflows
3.2 Using a System Agent
Click any System Agent
A popup will appear with an overview
Click Use Agent
You will be taken to DataPeak Chat, where it loads automatically
Follow the prompts to begin interacting with it
The agent will guide you through its steps conversationally.
3.3 Pinning System Agents
When you use a System Agent, it becomes pinned to the left side of Chat for quick access.
To unpin:
Select the agent
The agent description widow will open
Click Unpin
Note:
Pinning works only for System Agents, not for Custom Agents.
3.4 Settings & Permissions
Admins / Sub Admins
Can adjust triggers
Configure outputs
Modify parameters
Business users
Can run System Agents
Cannot modify them
4. When to Use Agentic AI
Use Agentic AI when you need to:
Automate repetitive processes
Build custom workflows
Run pre-built data cleaning or integration tools
Generate outputs (PDFs, files, summaries)
Query datasets without manual work
Process data across multiple branches
Enrich, monitor, or transform data