RAAs streamline daily workflows like:
- Filling web forms
- Moving data between spreadsheets and apps
- Sending automated emails
- Switching between desktop applications
- Extracting and processing data using AI
Types of Robotic Automation Agents
There are three types of RAAs you can create, each tailored to specific automation use cases:-
Web AI Agent
- Automates repetitive tasks across web applications.
- Use it to capture and replay actions like logging in, form submission, clicking buttons, etc.
-
Data Entry AI Agent
- Designed to import data from spreadsheets into any web app.
- You demonstrate just one row of entry, and the AI does the rest.
-
Create Your Own (from Scratch)
- A flexible automation builder for both web and desktop tasks.
- Choose from a wide range of steps like loops, app controls, file operations, API calls, and more.
Step-by-step: Creating a Robotic Automation AI Agent
1. Navigate to AI Agent Creation
- From the Supervity dashboard, click on the “Create AI Agent” button on the left sidebar.
2. Choose Robotic Automation AI Agent
- You’ll see different AI agent templates.
- Select “Robotic Automation AI Agent” from the list.
3. Choose Template or Start from Scratch
- Web AI Agent: Starts with web-based actions.
- Data Entry AI Agent: Asks for a spreadsheet and target web app.
- Scratch: Opens a blank canvas.
4. Name Your Agent and Add Steps
- Name your automation and click on Start to begin building the workflow.
- Use the ”+” button to add steps to the flow.
5. Build the Walkthrough
- You’ll be taken to the visual builder.
- Click the ➕ buttons to add steps in your flow.
- While building your walkthrough, each step can either:
- Add a Skill from your existing skill collections,
- Or Record your screen, where your actions are automatically captured and inserted as steps.
These recorded steps can later be reused as automations — making your co-browsing agent both instructional and functional.
6. Finalize and Publish
- Use the top options to Play, Save, Export, or Publish your Robotic Automation AI agent.
- Once published, the walkthrough can be triggered for users within your product.
Available Steps in RAA
Below are the key types of steps you can add when creating an agent from scratch:| Step Type | Purpose Example |
|---|---|
| Add a Skill | Reuse existing automation logic as a skill |
| Add Dataset | Add spreadsheet input to use in looping data entry |
| Loop | Repeat steps for each row or input |
| Show | Record desktop or web action using screen capture |
| Read/Write | Handle file, folder, or table actions |
| Clipboard | Copy/paste operations |
| Keyboard/Mouse | Automate interactions like keystrokes or clicks |
| Service Call | Call APIs or backend systems |
| Outlook Mail | Automate email actions via Outlook |
| Generative AI | Use GPT for dynamic content creation or extraction |
| RAG | Retrieval-Augmented Generation for query over knowledge |
| Try/Catch | Add error handling for robust workflows |
| Switch App | Toggle between open apps |
| Excel | Automate spreadsheet operations |
| Database Action | Read/write to a connected database |
| OCR | Use Supervity’s OCR to extract data from images/docs |
| S3 Actions | Download/upload files to/from Amazon S3 |
Integration
Once your Robotic Automation Agent is ready:- Click the three dots (⋮) on the agent card.
- Select Integration to get the script.
- You’ll see a JavaScript snippet that can be embedded into your web app to enable a conversational AI assistant based on this automation.
- You can even pass query parameters like chat background color, font size, etc.