Process management
Processes let you define steps for your AI Agent to follow so it can resolve specific customer inquiries. While some inquiries require your AI Agent to use exact wording and terminology (e.g., for compliance-based topics) to be resolved, others are more complex, requiring gathering information across multiple business systems, and dynamic decision-making logic based on that information.
For example, to refund a customer’s order, your AI Agent might have to first validate the items the customer wants to get a refund for, and ensure they’re covered in your business’ refund policy (e.g,. total amount and time since delivery), before executing an Action to process the refund.
You can use Processes to equip your AI Agent with the steps and order of operations to take to help your customers with their problems, without having to worry about scripting your AI Agent’s responses to customers’ questions - it’ll use that information to write responses dynamically.
How your AI Agent uses Processes
When you create a Process, you give your AI Agent guardrails around how to solve a customer inquiry. But rather than writing a script to execute, think of a Process as a description of what to do in a given scenario. Your AI Agent can then take that description to understand the steps to take and in what order, while also reasoning independently through how best to support the customer.
Here are some examples of some behavior you might notice when your AI Agent follows a Process:
-
“Has order AB87312 shipped yet?”
If your AI Agent needs an order ID to look up an order, it can extract the value from a previous question the customer asked.
-
“Actually, can you help me with something else?”
Based on new input from the customer, your AI Agent can decide to change course and help the customer with a different issue.
When your AI Agent is done executing all of the steps in a Process, it will use all of the information it gathered to write a response to the customer’s question. There’s no need for you to write its messages for it!
Understand how Actions and Processes work
Both Actions and Processes allow your AI Agent to assist customers by using business systems like APIs and getting required information from customers. Based on its own reasoning, your AI Agent still has the choice to use one or the other, but you can set it up for success by considering what kind of problem you want to solve.
-
An Action is an individual API request you can configure. For issues that can be solved with a single API request where the customer just needs the API response, an Action is a great way to quickly get that information to them.
-
Processes
Processes offer more flexibility than Actions. You can use Actions as building blocks inside a Process, so if you need to run multiple API requests in a specific order, you can create an Action for each of those API requests, and put both of those Actions in a Process.
You also have more flexibility with variables when you use Processes. If you need to save information gathered using an API request in a variable so you can use it in your future data reporting, save the variables in an Action.
For any Actions you want to use only in Processes, make sure you set them to Inactive. You can use any saved Action in a Process, but inactive Actions won’t run on their own, and won’t confuse your AI Agent when it’s deciding how to help a customer.
Understand how variables work in Processes
You can use variables as placeholders you can get from customers, like name or accountNumber. You can use the same variables between Actions and Processes - for example, in a single Process, you can use one Action to retrieve a customer’s shipping address and save it in a variable, then use another Action to ship an item to that address using that same variable.
But be careful when you’re using variables in multiple places! You can’t know what order Actions and Processes are going to happen in, and for scripted content like your AI Agent’s greeting, even if you save information in a variable, you can’t reuse those variables in an Action or a Process.
So if you need to use the same variable in multiple Actions or Processes, it’s best to configure each one to capture that variable. If your AI Agent already has a value assigned to a variable, it’ll know it doesn’t have to repeat the question to a customer.
Variables are color coded based on where the information they contain came from:
The different types of variables can come in handy in the following situations:
-
Store information for future data analysis
You can use variables to store information about the customer and conversation so you can analyze that data in the future. If you use any customer information in your reporting or analysis of Ada’s performance, make sure to store this information in a variable.
- If your Action’s output contains the customer’s tier or segment, store it in a global variable.
- If your Action’s input captures the customer’s email, store it in an auto-capture variable.
Then, you can use that data in two ways:
- Use the
variablesattribute in the conversation object in the Data API. The data you fetch from the API will contain variable information, which you can use to filter and analyze the conversation data. - On the Ada dashboard, filter reports by variable values.
-
Use the same value within a Process
You might need to use the output from one Action as the input for the next Action. For example, you might have an order lookup Action that outputs an order ID, and another Action that takes the order ID and looks up the order’s status. To ensure your AI Agent can successfully pass information from one Action to another, use variables.
-
Build dynamic logic into a Process
If the steps to resolve a customer’s inquiry are dynamic and change based on information gathered either directly from the customer or from an Action, you can use a Conditional block to provide different content depending on a variable’s value. For example:
- If the value in the customer’s
trial_periodvariable isless than30days, allow the customer to extend their trial - If the value in the customer’s
airport_originvariable isequal toYYZ, offer the customer a flight upgrade
- If the value in the customer’s
Track Process engagement with reporting
You can track how customers engage with your Processes by adding variables to capture specific engagement data, then using Ada’s reporting features to analyze that data.
Set up variables for Process tracking
To track engagement with a specific Process, add a variable at the top of the Process to capture when customers interact with it. For example, if you want to track customers who express a desire to cancel their service:
- In your Process, add a Set Variable block at the beginning of the Process flow.
- Create a variable like
cancellation_requestedand set its value totruewhen the Process is triggered. - This variable will be captured for every conversation where the Process runs, allowing you to track engagement.
View Process reporting data
Once you’ve set up variables to track Process engagement, you can analyze this data across multiple locations in Ada:
Reports: All performance reports support variable filtering
- On the Ada dashboard, go to Analytics, then open the Reports tab.
- Open any report (Conversations breakdown, Performance summary, etc.).
- Use the Filter by variable option to view only conversations that include your Process tracking variables.
Homepage Dashboard: Filter dashboard analytics by variables
- On the Ada dashboard, use the variable filters to analyze Process engagement data.
- View metrics and trends filtered by your Process tracking variables.
Convos View: Search and filter individual conversations
- Go to Convos to view individual customer interactions.
- Use variable filters to find conversations where specific Processes were engaged.
For each variable filter, you can:
- Filter by whether the variable Is Set or Is Not Set
For example, to see the number of engaged conversations where customers expressed a desire to cancel, filter by cancellation_requested = true across any of these locations. This will show you how many customers engaged with your cancellation Process.
For more information about using reports and variable filtering, see Performance reports.
Configure a Process
-
On the Ada dashboard, go to Config > AI AGENT > Processes.
- To create a new Process, click New Process. A page to build your new Process on opens.
- To create a Process using an existing Process as a template, open the Process you want to duplicate. Then, at the top of the page, click the More options button
, then click Duplicate Process. Your duplicated Process opens, which you can then edit as required.
-
Under Identify this Process, enter a Name and Description for the Process.
The Name and Description provide important context to your AI Agent about when it should trigger the Process. For the description, write it as if you’re explaining to a human agent the situations in which the Process is appropriate to use.
-
If required, you can restrict Ada articles to certain users, based on information your AI Agent collects about your users and saves in variables.
You can only use variables your AI Agent can collect through your browser, or that you collect in a block and allow to be available outside of the structured content the block is in. You can’t use variables your AI Agent collects using Actions.
Under Set availability, choose which customers you want the Process to apply to:
-
To make the Processavailable to all users, select Everyone.
-
To restrict the Process to certain users, select Based on the following rules. A section expands where you can enter the logic your AI Agent will use to decide whether to serve the Processes.
- Under Where, in the Choose a variable list, select a variable.
- In the next dropdown, select an operator so you can define a relationship between the variable and the value you want to target.
- In the Value field, enter or select a value for the variable that you want to use to target users.
- If required, add additional conditions.
- To add a new top-level condition, click Add
.
If you’re adding your first additional top-level condition, in the dropdown that appears, choose And or Or as the operator for all of your top-level conditions.
- To create or add to a group of conditions, click the Add to group icon
beside a condition you want to include in the group.
If you’re creating a group, in the dropdown that appears, choose And or Or as the operator for all of the conditions in that group.
Understand comparison operators
Comparison operators are logic statements that tell your AI Agent to match customer information that’s captured in the variable you’re using. The available operators vary based on the variable type you’re using:
-
-
Under Build the Process, put together the sequence of events that you want to occur in your Process. Your AI Agent will follow the blocks in order from top to bottom each time it decides to trigger this Process.
Click a block name below to learn more:
- To save your new Process as a draft, click Save. If you save it as a draft, it won’t be available to your customers until you publish it.
- Optionally, you can test your Process, either by clicking Test process to test it in isolation, or by clicking Test AI Agent to also test whether it gets triggered when you want it to.
- To publish your Process, beside the Inactive toggle at the top of the page, click the toggle so it says Active. As soon as you do, your AI Agent starts using your new Process with customers.
Understand block support by channel
Not all blocks that are available in Processes are supported across all channels.
After creating a Process, when you click Save you will receive a warning message if your Process contains a block or blocks that are not supported for a given channel and it is currently available in that channel.
In order to resolve these issues you can either:
- Remove any unsupported blocks from your Process
- Set availability for the Process to only channels where all blocks in the Process are supported
- Use the Conditional block to prevent specific blocks in a Process from being used in channels where they are not supported
Edit or delete a Process
After you or a colleague create a Process, you can go back to it so you can edit or delete it.
- On the Ada dashboard, go to Config > AI AGENT > Processes, then open the Process you want to delete.
- Make your required changes to the Process.
- You can edit the existing Process and then click Save to update it and keep it active.
- You can turn the Active toggle at the top of the page on or off to control whether the Process is accessible to customers.
- You can delete the Process by clicking the More options button
at the top of the page, then clicking Delete Process.
View and manage previous versions of a Process
Your AI Agent saves a history of all edits you’ve made to your Processes over the last 30 days. You can revert your changes to a Process back to any of these saved versions.
- On the Ada dashboard, go to Config > AI AGENT > Processes, then open the Process you want to manage versions for.
- At the top of the page, beside the version icon
, you can see when the Process was last edited. Click the icon or the text to open a pane that contains a list of versions of the Process over the last 30 days. You can click a version to see the Process at that time. - Optionally, for significant versions, you can add a name and description for future edits. To do this, hover over the version, click the More options button
that appears, filling in the Version Title and Description fields as required, then clicking Save. - To revert to a previous version of your Process, select the version you want to go back to. Then, at the top of the page, click Restore Version. Review your changes to ensure you’re not reverting anything you want to keep, then click Save to save your reverted changes.