What Are Agents?
Agents in BonData are automated workflows that let you do something useful with data from your connected systems. Each Agents defines:
Filters (Conditions): Criteria that identify which records need attention (for example, “Records missing a required field”).
Actions: What BonData should do once those conditions are met—such as creating or updating records, generating a CSV or Google Sheet, notifying a Slack channel, or sending an HTTP request.
Bonds:Shows AI-generated recommendations for potential connections between
systems.Once confirmed, these connections help unify your data across different platforms,ensuring seamless workflows and automation.
Bonds specify which fields from different systems are linked.
By relying on BonData to monitor and act on your data, you avoid manually juggling multiple apps or triggers.
Why Use Agents ?
Automate Repetitive Tasks: Let BonData handle data updates, generate exports, or trigger notifications—freeing you from manual effort.
Single Point of Control: Combine data and actions across systems (Salesforce, a custom database, Slack, or a data warehouse) without building custom integrations.
Flexible Execution: Choose whether tasks run automatically or require your approval first—so you stay in control while still reaping the benefits of automation.
Key Concepts
Integration: Any external system or data source you connect to BonData (e.g., Salesforce, a CSV file, internal APIs, etc.).
Filters (Conditions): Specify when an Agent should act (e.g., “Lead record missing a primary contact email”).
Actions: Tasks BonData performs, such as:
Create a Record (in Salesforce, for instance)
Update a Record (change status in a connected database)
Notify on Slack (ping #ops-team about critical data)
Generate CSV / Google Sheet (compile a list of matching records for offline review)
Send HTTP Request (push data to a custom endpoint)
Tasks: Each Action for every record that meets your Conditions is a separate “Task.” For instance, if 20 records need updating, BonData generates 20 tasks.
Automatic vs. Manual Execution:
Automatic: Tasks run immediately when the Agent condition is satisfied.
Manual: Tasks are queued and only execute if/when you confirm them individually.
Creating Your First Agent
Access Agents
From the BonData dashboard, open the Agents section.
Click Create New Agent.
Name and Describe
Give your Agent a name that reflects its purpose (e.g., “Data Cleanup for Missing Emails”).
Optionally add a brief description.
Set Filters
Choose which system(s) or fields you want to monitor.
Define the logic (e.g., “Accounts with no assigned owner”).
Combine multiple criteria with “AND” or “OR,” as needed.
Field comparison: allows you to create a condition that compares between different fields (e.g. when "Company Name" in Hubspot Not Equals to "Account Name" in Salesforce)
Filter insights: to the right side of the filter, you can click to see the amount of records that currently match the filter (condition) applied. In order to see the insights, you should save the Agent first.
Choose Actions
Decide what BonData should do when it finds records matching your conditions. Common examples:
Create Record in Salesforce
Update Record in your warehouse
Notify a Slack channel
Generate CSV/Google Sheet
Send HTTP Request to an external service
Decide on Execution Mode
Automatic Execution: Each matching record triggers tasks right away.
Manual Execution: Tasks appear in your queue, allowing you to approve or skip them individually.
Activate the Agents
Once you’ve finalized the conditions and actions, click Activate.
BonData starts monitoring new (and existing) records on your defined schedule.
Mentions: Mapping Fields & Translating Values
When creating or updating records in an Action, you often want to pull values dynamically from one source—like a CSV file or a connected system—and insert them into fields in another system. Mentions let you do exactly that.
What Is a Mention?
A Mention is a placeholder you insert in an Action to dynamically reference a field or column from another data source. For example, if you’re importing records from a CSV into Salesforce, you can “mention” the CSV’s columns to tell BonData which values go where.
Basic Mentions
Click Mention within the Action configuration.
Select the relevant integration (e.g., “Salesforce”).
Choose the specific object (e.g., "Account").
Choose the specific field (e.g., “Account Name”).
BonData will insert the Salesforce “Account Name” field into the action that you want to take.
Use Case Example:
CSV columns: Account Name, Region, Industry.
Action: Create a Salesforce Account.
Under “Name,” insert a mention of Account Name.
Under “BillingState,” mention Region.
Each row in the CSV triggers a new Salesforce account creation, using the row’s column values.
Advanced Mentions & Conditional Translations
You can also transform a value from the source before it populates another system:
Scenario: Salesforce’s “Opted Out of Emails” field needs “Yes” or “No,” but your CSV column can be 0, 1, or Null.
Solution:
If CSV “OptedOut” = 1 → mention “Yes” in Salesforce.
Else → mention “No.”
If a CSV field is blank, you can also provide a static default (“No”) to ensure the target field is never left empty.
Agents Navigation: Filter Insights, Planner, and Trigger Settings
When you open or edit an Agent, look at the top-left navigation area to find three important tools:
View Filter Insights (Rightmost Button)
Purpose: Quickly see how many records match the filters you defined.
Example: If your condition is “Accounts with critical support issues,” Filter Insights might show “10 records match.” This helps you verify your filter is correct before (or during) activation.
Trigger Settings (Leftmost Button)
Trigger Behavior:
Change (default): Tasks are generated only for newly created or newly changed records that meet your filters.
Always: Each time the Agent runs, it will create tasks for all records that match the filter, whether they are new or not.
Example: If 20 accounts match “critical issues” on Monday, a daily alert on “Always” mode might show all 20 each day. If a new account is added on Tuesday that also has critical issues, you’ll see 21 on Tuesday’s alert.
If you prefer only to see newly added matches each day, stick with Change mode.
Run Frequency:
By default, BonData checks your conditions every 15 minutes (relevant only when the Agent is set to run "automatically", which is off by default).
You can customize this to any interval, including a cron expression.
Remember that time is always in UTC.
Use these three tools to fine-tune how your Agent identifies matching data, how it navigates your systems’ connections, and how frequently tasks should be generated.
Save / Test / Run in BonData
Save your workflow: Click "Save" to save your changes.
Test : Lets you simulate the Agent before turning it on. It helps verify that each step works as expected, without actually triggering live tasks (like sending real emails or updating data).
Run : To Activate your Agents : Return to the workflows overview and turn the workflow on. It will now automatically run according to its trigger.
Before Running your Agent
When creating a new Agent in BonData, the system temporarily stores relevant records solely for Agent execution purposes.
Due to this temporary storage process, there may be a brief waiting period between creating your Agent and its first execution. Before your Agent can run, both metadata (structure of the fields) and actual data need to be successfully fetched into BonData.
To read more about how you can check the load status of the data, or trigger a manual load if new data was just added, click here:
Frequently Asked Questions
Can I add multiple actions in one Agent ?
Absolutely. Each record that meets your conditions can trigger multiple tasks.
What if I only want to suggest actions but not apply them automatically?
Select manual in your Agent settings so tasks require confirmation.
Can a single Agent work across multiple data sources?
Yes. Conditions and Mentions can reference multiple integrations.
How do I revert a change if something went wrong?
Check the Task logs to see which records were updated, then revert those changes in the integrated system. BonData doesn’t store your entire data set, so rollbacks happen externally.
Best Practices
Start Simple: Test a single condition and action, referencing one field at a time.
Scale Up Gradually: Once confident, add more columns or advanced mention logic.
Audit Agent Periodically: Business rules or data fields may change. Keep your filters and mentions current.
Monitor Task Volumes: An unexpected surge might indicate overly broad filters or incorrect mention logic.