Kognitos is an automation platform for building and running end-to-end business processes using plain English. Discover how it works and get started below.

Learn More

🔍 Explore the Docs

Kognitos is an automation platform that enables users to automate end-to-end business processes in plain English. It’s designed to manage complex tasks, including those with many exceptions, and improves over time based on user feedback.

Why Kognitos?

Traditional automation tools require months of training and only work for simple processes with few exceptions. However, businesses need tools that can handle complexity and change. Kognitos is built to be easy to use and requires minimal training. You can create automations, manage exceptions, and review logs—all in plain English.

What is a Comment?

A comment is a line in an automation that is skipped during execution. Comments are used to add notes and annotations. Additionally, they can be used to disable a part of your automation temporarily.

How to Add a Comment

To add a comment, use the # symbol at the beginning of a line. For example:

⚠️ Single-Line Comments Only

Currently, the platform only supports single-line comments. If you want to add multiple lines of comments, you’ll need to use the # symbol on each line. For example:

Within the Guidance Center, you may see Suggestions listed below an exception. These suggestions represent learning techniques provided by the system as potential solutions, ranked by how closely the current document matches the document where the learning technique was previously applied.

How to Apply Learning Suggestions

1. Hover over a Technique to see the confidence level displayed as a percentage. This score measures how similar two documents are.
2. Hover over the top right corner of the Technique and click on the Play button to open a Mini-Playground. This is where you can test the learning without affecting the main process run.

3. If the result is as expected, click Next and then click Use this method to apply the learning. If the result is not as expected, you can modify the technique in the Mini-Playground to get the correct value.

What is the Debugger?

The debugger is designed to simplify the debugging process. It allows you to add breakpoints to start and pause your automation as needed, enhancing your development workflow and experience.

Overview

Invoking a subprocess is one of the three ways a parent process can call a subprocess in Kognitos. When a subprocess is invoked using the invoke keyword, the parent process does not manage its execution or output.

Overview

A loop allows you to repeat actions for multiple items. In Kognitos, you can create a loop using process each x as follows, where x represents the item in a set. Loops are especially useful for batch processing, where you need to handle multiple items in a consistent manner.

Overview

Learnings are rules that guide your automation’s decision-making. These rules help your automation interpret and respond to different scenarios based on past interactions and defined logic. Learnings can range from simple data references, such as:

To more complex rules, like:

Overview

Scheduling a process enables you to automate processes to run at set intervals—such as hourly, daily, or weekly. This ensures tasks are completed consistently without manual input, improving efficiency by minimizing the need for oversight. This feature is particularly useful for time-sensitive or repetitive tasks, enabling users to focus on more critical work while Kognitos takes care of background processes.

Kognitos allows your automation to make decisions based on logic. The keywords if, then, and else help control your automation's behavior in different scenarios, enabling dynamic workflows.

if / then Syntax

The basic syntax for a conditional decision is:

Example

if / then / else Syntax

To add more flexibility, you can use an else statement to specify an alternative action when the condition is not met:

Example

Overview

One of the three ways a parent process can call a subprocess in Kognitos is by using the Process Run Widget. In this setup, the parent process controls the execution and output of the subprocess.

How to Call a Subprocess Using the Process Run Widget

1. In your automation, write the keyword run followed by a slash /

2. From the menu that appears, select Process.

3. Click on the Process drop-down menu to select the process you want to call.

4. Click Save.

Passing Data to a Subprocess

Passing data to a subprocess allows the parent process to provide the specific inputs the subprocess needs to do its work. This makes subprocesses more dynamic and reusable, as they don’t depend on hardcoded values. To enable this, the parent process must explicitly pass the required data, and the subprocess must be designed to accept and return values accordingly.

Parent Process

In your parent process, add a new line after run <subprocess>. Indent and reference the data needed for the subprocess run on the new line. If a data element is required but not explicitly provided, Kognitos will raise a .

Subprocess

In your subprocess, include one of these lines at the end of your automation to return data to the parent process. This syntax informs Kognitos of the subprocess's result so it can be accessed in the parent process.

• For a single result: the result is {result}

• For multiple results: the results are {result 1}, {result 2}, ...

Examples

1. Returning a Single Result

This example shows how a subprocess can be used to calculate a total. The main process, Process an Invoice, calls the subprocess calculate the total and provides the price. The subprocess uses the price to calculate and return the total.

2. Returning Multiple Results

This example shows how to return multiple results from a subprocess. Here, the subprocess Split Full Name returns two results: the first name and the last name .

Overview

In Kognitos, exceptions occur when an automation cannot proceed without additional context or clarification. Instead of making an assumption that could lead to errors, the system raises a question (also called an exception) and routes it to a human for review.

This human-in-the-loop design ensures that business processes remain accurate, reliable, and trustworthy. Exceptions are not system failures — they are intentional checkpoints that keep automations accurate and reliable.

Why Do Exceptions Occur?

Exceptions can arise in Kognitos for a variety of reasons. The most common include:

Ambiguity

Multiple possible interpretations are possible, and the system cannot determine which is the correct one without user input.

Example: An email approval workflow asks, “Approved?” but doesn’t specify if it refers to cost, vendor, or timeline, so an exception is raised for review.

Missing Information

A required data input is not available, preventing the automation from completing the step.

Example: While processing an employee expense report, the system detects that a hotel receipt is missing the check-in date, so it raises an exception for confirmation.

Document Variability

The information exists in the document, but appears in an unexpected format or structure. This makes it difficult to extract with certainty.

Example: An invoice shows the total amount as “$5,000 (including VAT)” in a note at the bottom instead of in the standard “Total” field, so the system raises an exception for review.

Business Rules

A step requires human decision-making, often due to risk, compliance, or policy considerations.

Example: While reviewing a contract, the automation extracts a clause about early termination fees. Because company policy requires legal review of any termination penalties above $10,000, the system raises an exception so a human can confirm how to proceed.

How to Resolve Exceptions

Exceptions in Kognitos are resolved by providing guidance — giving the system the input it needs to understand how to proceed. This can include clarifying ambiguous values, pointing to the correct information in a document, or confirming a decision.

The makes this process efficient by allowing you to review open exceptions across automations in one place and provide the necessary clarification or correction. Once guidance is provided, the automation resumes where it left off.

Guidance Examples:

• Identifying the correct field in a scanned invoice that represents the “Total Amount Due.”

• Selecting the correct shipping address from a document that lists multiple locations.

• Confirming the contract start date when it appears in narrative text instead of a standard date field (e.g., “services commence on June 1st”).

• Verifying the correct vendor name when variations or abbreviations appear across different documents (e.g., “Northstar Technologies” vs. “Northstar Tech LLC”).

Summary

Exceptions in Kognitos are not errors; they are intentional checkpoints that safeguard accuracy. By providing guidance in the , you ensure that automations continue smoothly, learn from your input, and become more resilient over time.

Creating a Process

After testing your automation in the Playground, the next step is to save it as a draft process. A draft process allows you to formalize and iterate on your automation before publishing it to production.

Publishing

When your draft is ready, you can publish it to make it production-ready. Publishing locks the process to prevent further edits, ensuring consistency across all runs. To publish your process, click on Publish Process.

Editing

Once a process is created, you can edit it to make improvements or updates. Edits are made in a draft state, allowing you to test changes before publishing them live.

Menu

Access the widget menu by typing / when editing a playground or process automation. A dropdown will appear, allowing you to select from the available widgets.

Overview

When reviewing automation runs in Kognitos, you may want to scan for specific data points important to a process, such as an invoice number, payment status, or user ID. Starring Data allows you to pin key data elements so they stand out across process runs and are easier to track.

Overview

Kognitos provides a self-service console for creating and managing API keys. Keys can be scoped to an entire organization or a specific agent, depending on the level of access required.

Overview

Context Based Learning (CBL) is a powerful feature that translates key information in documents (ex: field names, column names, and large words) into numerical values. These values can then be compared between two documents to determine their similarity.

How Is It Useful?

You don't have to trigger every automation run manually. Kognitos lets you trigger automations by sending an email — a powerful way to automate tasks like onboarding or invoice processing.

How Email Triggers Work

A unique email address is generated for every Kognitos process. When someone sends an email to this address, it automatically starts a new process run and passes the email contents to the automation, including:

• Subject line

• Sender address

• Body (plain text or HTML)

• Attachments (files, spreadsheets, images, etc.)

How to Configure Email Triggers for a Process

How to Trigger a Process Run

Once you've configured email triggers for your process, anyone with the proper permissions can start a process run by sending an email. Here's how:

1. Compose a new email in your email client (Gmail, Outlook, Apple Mail, etc.)

2. Set the recipient to the unique email trigger address you copied during configuration.

3. Write your email with any relevant information: subject line, email body, and attachments.

4. Send the email — this immediately triggers a process run in Kognitos.

As soon as your email is delivered, Kognitos automatically starts a run of that process. Your automation can access all parts of the email (subject, body, attachments, sender info).

Troubleshooting

If your process isn't starting when emails are sent:

1. Ensure the sender has adequate permissions to send emails that trigger the process.

2. Double-check the trigger email address — did the email go to the correct address?

3. Check if the email delivered successfully. If the email fails to deliver, the process won't trigger.

4. If you're using an email forwarding rule, verify that it's active and correctly configured.

Best Practices

Managing Trigger Permissions

Use the most restrictive permission setting that meets your needs:

• Collaborators: Best for internal-only workflows.

• Restricted: Ideal when only a specific set of users should be able to trigger the process.

• Anyone: Use with caution & ensure the automation includes validation steps to prevent misuse.

Validate Incoming Emails

Add logic in your automation to validate the email content, such as checking for specific subject lines, senders, or attachments. For example, to access the attachments of an email that triggers a process, we would write:

This statement grabs each attachment from the email in order to take further action. You can also use the following phrases to get details about the email to use in your automation:

• get the text

• get the sender

• get the recipients

Overview

Batch processing is a method of handling data by dividing it into smaller groups (batches) and processing each group separately. This is useful when working with large volumes of information such as customer records, invoices, or files as it improves efficiency and error handling.

How It Works

When you create batches, the items are divided into groups of the specified batch size. Each batch becomes a separate collection that you can process independently.

Example: Instead of sending 1,000 customer emails all at once, you can create batches of 50 emails each, resulting in 20 manageable batches to process one at a time.

Syntax

Below is a line-by-line overview of the automation syntax. Expand each line to learn more.

Examples

1. Basic Batch Creation

The following example shows how to create batches from a simple list of numbers:

2. Processing Customer Records

Here, the customers are processed in batches of 10.

3. Invoice Processing

In this example, the table's rows are broken up in batches. Then, each batch is processed in parallel using .

Overview

Data validation ensures that a process’ data is appropriate and logical, allowing it to be trusted for decision-making. In Kognitos, you can use the ensure keyword to specify conditions for data validation. When data cannot be validated, the automation will raise a question, allowing you to review, edit, and confirm values in a single location.

Writing Validation Logic

ensure

The ensure keyword is used within a validation statement that:

• Begins with ensure

• Is indented within validate the following

• Uses an operation that:

validate the following

The phrase validate the following groups multiple validation statements together. When a validation fails, the data from all statements will be aggregated into a single view.

Examples

1. Validating Text Data

2. Validating Numeric Data

3. Validating Multiple Data Elements

Evaluation

Each validation statement evaluates to Condition Met (true) or Condition Not Met (false).

The automation proceeds when all validation statements have a Condition Met. However, if any statement has a Condition Not Met, the system raises a question, requiring you to provide guidance to proceed.

Skipped Statements

A validation statement is marked as Skipped when the syntax is invalid. To avoid this, ensure the operation is a valid boolean operation supported in Kognitos.

For example, the statement above is skipped because the use of "has" is not supported in a Kognitos operation. Instead, the validation statement could be written using the contains keyword: ensure the vegetables contains "broccoli".

Providing Guidance

When a validation statement within validate the following has a Condition Not Met, a question is asked to Please review the following facts.

Reviewing Data

Data is listed for review in the same order it appears in the validation statements. For conditions that are not met, a message will appear below the data value in red, stating the failed validation.

Manually Editing

You can manually edit and override data values as needed.

Finalizing Your Review

Confirm your review by checking the box labeled "I have reviewed all values and they are accurate". Then, click Submit.

Spot Check Percentage

The spot check percentage controls the likelihood of a question being raised when all the validation statements evaluate to true. The percentage does not represent accuracy or error rate; it controls the probability of triggering a question.

Syntax

Behavior

• If any validation statement fails, a question is always raised.

• If all validation statements pass, a question is raised based on the spot check percentage.

  • Example: If the spot check percentage is 30, then there is a 30% chance that a question is raised when all the validation statements evaluate to true.

Example

In this example, there is a 50% chance that a question is still raised for review, since both validation statements evaluate to true.

Overview

In Kognitos, one process can call and run another process. The called process is referred to as a subprocess. There are many advantages of using subprocesses:

• Modularity: Decomposing automations into smaller processes enhances structure and organization.

• Readability: Well-organized automations are easier to read and understand.

• Scalability: Automations can be scaled to handle multiple tasks.

• Maintainability: Isolated processes simplify debugging and updating individual components.

Mechanisms

There are several ways to call another process in Kognitos:

1. : This method is ideal for single-run scenarios where you need to capture the result and status of a subprocess.

2. : This method is useful for scenarios where the parent process can continue executing without waiting for the subprocess's status or completion.

3. : This method provides control over parallel execution when handling multiple subprocesses simultaneously.

Comparison

Agent

An agent in Kognitos is a dedicated execution environment where your automations live and run. Each agent is equipped with its own set of learned knowledge, configurations, and connections to other tools, allowing it to operate independently and securely.

Automation

An automation is a series of instructions that tells Kognitos what to do. It can refer to either a draft workflow in the or a fully developed and reusable .

Book

A is a set of related automation . Books can be added to an agent to expand automation functionality and integrate with external tools. Kognitos offers a library of standard Books for common skills and integrations, including Microsoft Excel, Salesforce, and Slack.

Book Development Kit (BDK)

The is a software toolkit for developing custom books in Kognitos. Using the BDK, developers can write code that translates plain English procedures in Kognitos to actions that a specific third-party application can understand.

Exception

An exception (question) arises when the system needs additional clarification to proceed with an automation. Exceptions are resolved by providing user guidance in the .

Guidance Center

The Guidance Center is where you manage all in one place. When an automation encounters an issue or requires additional context, it raises a for your input. Here, you can view and resolve exceptions across all processes in your agent.

History

Kognitos tracks major changes to an Agent, allowing users to review and restore past versions through the History tab. You can click on any of the options displayed in the History tab to view the Agent at that point in time.

Kognitos AI Assistant

The Kognitos AI Assistant (KAIA) is used to answer questions, find guidance, or get in touch with our support team. You can use this tool to get help anytime by accessing it in the bottom left corner of the platform.

Learnings

Learnings are rules that guide your automation’s decision-making. When you resolve an , you can teach the agent how to handle similar situations in the future. These learned responses are stored as "learnings, " allowing the automation to become more intelligent and autonomous over time.

Playground

The Playground is a dynamic development environment where you can create, test, and refine your automation workflows. It provides a flexible space to write procedures in plain English, run them to see the immediate results, and debug any issues before creating a .

Procedure

A procedure is a single automation task defined in plain English. Each procedure describes a specific action and follows a defined grammatical syntax that Kognitos understands, such as open a file, extract data, or send an email. These procedures are the building blocks for Kognitos automations.

Process

A process is an automation workflow built using one or more . It represents a complete business task, such as processing an invoice from receipt to payment.

Processes can be in a Draft state during development and testing, or in a Published state that is production-ready. Each version of a process is tracked, allowing for clear change management.

Run

A run is a single execution of an automation. All automations, whether a Playground, a Draft Process, or a Published Process, can be run manually. However, only published Processes support automatic triggers like email, API, or scheduled runs. Every time a process is triggered, either manually or by a scheduled event, it creates a new run.

Subprocess

A subprocess is a process that is called by another process. Learn more .

Overview

Kognitos supports the ability to export automation agents from one environment and import them into another. This allows you to move published automation content across agents in a controlled way.

Exporting an Agent

Exporting lets you download a snapshot of an agent’s published state. The export includes only published processes, learnings, and books — draft processes, playgrounds, and configuration credentials are not exported. Anyone with access to the agent can export it.

How to Export an Agent

Importing an Agent

You can import an agent using a .json file generated from the export process. There are two ways to do this — into a new agent or into an existing agent. All imported items are added as drafts. The agent's published content is not affected unless you review and publish the imported content.

1. Import into a New Agent

Use this option to create a new agent.

2. Import into an Existing Agent

This option allows you to update an existing agent and is available only to agent admins.

How to Import an Agent

Overview

The ask keyword allows automations to prompt the user with a custom question. These prompts are raised as custom exceptions during execution.

Syntax

Write the keyword ask followed by your question in quotation marks. When raising an exception, Kognitos will use this text for the question:

Optionally, you can provide a set of answer choices. Kognitos will display these options in a dropdown menu where the user can select a value for guidance.

Examples

Without Choices

Specifying Choices

Example 1

In this example, the automation will raise an exception asking What is the notification method? and the user will be presented a list of dropdown choices including Email, SMS, and Push Notification.

Example 2

Here, the automation will raise an exception asking What is the preferred language? The user will be presented with a list of dropdown choices that contain English, Spanish, Hindi, and French.

Providing Guidance

When Kognitos raises a custom exception with the keyword, you’ll need to guide the automation by choosing how to respond. The resolution methods depend on whether the exception includes choices.

When Choices Are Provided

In this case, the recommended resolution method is to Pick a Suggested Value. This will allow the user to select one of the predefined options from the dropdown list. This is the safest and most reliable way to resolve the exception.

When Choices Are Not Provided

1. Write in an Answer

Type a value directly into the text box for the requested field.

2. No Value

Indicate that no value should be used for now. A dash (-) will be inserted as a placeholder. This is best used when the information is unavailable or not relevant.

3. Skip this Step

Bypass the current step entirely.

4. Compute an Answer

Modify the automation syntax to avoid this exception. Use the Mini-Playground to test changes safely before applying them.

Overview

Exception routing is a process configuration setting. It allows agent admins to configure assignment and notification rules for exceptions. When a process encounters an (a question requiring guidance), these rules ensure that the right person is automatically assigned to handle it and that the appropriate team members are notified.

Overview

Bring Your Own Model (BYOM) provides a secure way to add, connect, and manage API credentials for your preferred Large Language Model (LLM) providers. By default, Kognitos uses OpenAI and Gemini, but BYOM allows you to configure alternative or custom providers at the agent level.

Overview

In Kognitos, defining data refers to naming a specific piece of information, allowing it to be stored, referenced, updated, and passed between steps within an automation.

Every data element (also referred to as a fact) consists of two parts:

1. A that identifies and labels the data.

Overview

Kognitos empowers you to write automations in plain English, turning natural-language instructions into executable workflows. This guide will walk you through core concepts, the development lifecycle, and the basics of writing an automation.

Core Concepts

Let’s kick things off with two important ideas: and .

Agents

An agent is a dedicated execution environment. Each agent comes with its own skills and settings, enabling it to operate independently. Agents are created to handle specific business tasks and can contain one or more to complete them.

Automations

An automation is a set of instructions that tells Kognitos what to do — process data, work with documents, send emails, call other processes, and more. These are workflows that can take the form of an early idea (playground) or a complete process.

Automation Lifecycle

Automations come in different forms as they evolve from early ideas into production-ready processes, following this lifecycle:

Start in the Playground

Start in the Playground — your initial workspace for writing and testing automations.

Create a New Playground

Create a new Playground by navigating to Playground , then click on + New Playground.

Write Your Automation

The input area of the editor is where you can define your automation workflow.

Writing Automations

Kognitos automations are written in English. They are composed of small, readable steps called procedures. By combining multiple procedures, you can form a complete automation workflow.

What is a Procedure?

A procedure is a single automation task. Each procedure describes a specific action and follows a defined grammatical syntax that Kognitos understands, such as , , or . Procedures are the building blocks for Kognitos automations.

Building a Workflow

Combine procedures to create an end-to-end automation. For example, you might extract fields from a receipt, summarize them in a table, and then email the result:

Auto-Write (HAL)

With Kognitos’ Hyperautomation Lifecycle (HAL), you can describe what you want in plain English and let the platform auto-write the automation for you. No need to worry about syntax or structure; write naturally (e.g., “extract date and total from a receipt and email me a summary”).

HAL interprets your intent, selects the right actions, and produces a draft workflow that you can run or edit as needed, allowing you to move from idea to execution faster.

Language & Logic

Beyond procedures, use , , , and other elements to write and structure automations in Kognitos. Learn more below:

Books

A is a set of related automation procedures (actions) that can extend an agent’s capabilities. Adding a book to an agent enables the agent to learn new skills and integrate with external tools. Kognitos offers many Books that can be added to your agents, which are built around either:

• A specific domain (Document Processing, HTML)

• An integration (Microsoft Excel, Slack, Salesforce, etc.)

1. Could not get a file as a scanned document

This exception occurs when the system fails to access a file due to formatting, permissions, or network errors.

2. Could not (something)

This exception occurs when the system is unable to complete a specific action. The exception message starts with "Could not" and describes the action that could not be completed. For example:

• Could not ask koncierge

• Could not dump a file

• Could not extract a table

3. Please provide (something)

This exception occurs when the system cannot find a match for a specified field. The exception message starts with "Please provide" and describes the required data.

4. The key does not exist. How would you like to proceed?

This exception occurs when attempting to access a key missing in the current context or data structure.

5. I don't know how to (do something)

This exception occurs when the system doesn’t know how to proceed. For example:

• I don't know how to convert 'ADFB00' into a number

• I don't know how to process this billing code

• I don't know how to translate this to a string

It usually occurs due to one of the following issues:

1. The agent doesn’t have the skills needed to perform the task (a missing Book).

2. The automation syntax is incorrect.

3. The data is in the wrong format or type.

6. Which procedure should I use?

This exception occurs when there are multiple procedures with a similar syntax and the agent needs clarification on which one to use.

7. The automation crashed in a previous run. Click 'Yes' if it's safe to continue

This exception occurs when an automation previously crashed during execution, and the system needs confirmation that it’s safe to resume.

8. Run had crashed. Do you want to restart?

This exception occurs when a previous run has crashed, and the system needs confirmation to restart the automation.

Overview

A run represents a single execution of a . Every time a process is ran, Kognitos captures all the details of its execution (inputs, outputs, and state changes). Each run is also assigned a unique ID, making them easy to track, monitor, and manage.

Overview

The Test Suite empowers you to test and validate any changes made to your processes. After modifying a process, run the Test Suite to compare the new results against expected outcomes. This makes it easy to assess the impact of your changes and ensure everything still works as intended.

How It Works

Test changes made to your processes with this 3-step process:

Getting Started

Get started by creating test cases. To do this, you'll need to add test cases and identify output.

1. Add Test Cases

A test case is a process run that serves as the baseline for how a process should perform. Test cases establish the standard against which future runs are compared. They can be added from existing runs directly within the Test Suite or from individual process runs.

2. Identify Output

The next step is to define what a successful outcome looks like for your test case. This is done by — highlighting important data elements in a process run. By doing so, you tell the Test Suite which output (ex: total cost, destination, payment status) should be validated to ensure a process runs as expected.

Running Tests

Once you've added test cases and identified output, you're ready to run tests. During a test run, the Test Suite runs the current version of each selected process for every test case. Then, it compares the output generated by the runs against the expected output in your test cases.

Starting a Test Run

Follow these steps to start a test run:

Stopping a Test Run

To stop a test run in progress, click Stop Test Run on the Test Run page.

Reviewing Test Run Results

All your test runs are summarized in the Test Runs tab of the Test Suite. Access the results of a specific test run by clicking on it from this overview page.

Overview

Here's a visual guide to reviewing your test run results:

Test Run Metrics

Displayed at the top of a Test Run are the following metrics:

• Passed Runs: The number of runs where all the output matches the expected results.

• Failed Runs: The number of runs where one or more outputs does not match the expected results.

• Guidance Required: The number of runs that paused during execution and required user guidance to proceed. These runs are neither passed nor failed until guidance is provided.

Process Breakdown

Below the overall summary on the Test Run page is a list of processes included in the test run. For each process, you can see a summary of the overall status, pass percentage, and run metrics. Click into a process to drill down into individual test cases.

Test Case Breakdown

Once you've clicked into an individual process within a test run, you'll see a breakdown by test case. Click on any test case to view a side-by-side comparison between the expected output and the test run output.

Addressing Test Failures

Test failures generally occur in one of the following scenarios:

Output Mismatch

The test fails because the expected output doesn't match the current output from your process. To address this issue, you need to analyze the discrepancy and ask: is the new output correct?

• If no, there is an issue with your process. Fix the automation logic causing the incorrect result.

• If yes, the test case is outdated. Remove the test case and add a new one that contains the updated output.

Output Not Found

The expected output is missing entirely, pointing to an issue with the process itself. To address this, update your process to ensure it generates the expected output. This involves debugging your automation workflow and logic to understand why the data isn't being generated.

Removing Test Cases

Remove invalid or outdated test cases from your Test Suite to keep it up-to-date. Test cases can be removed directly from the Test Suite or from process runs.

Remove Test Cases from the Test Suite

1. From the left navigation menu, go to the Test Suite.

2. Click on the Test Cases tab.

3. Click on a process to see its associated test cases.

4. Select one or more test cases to remove.

Remove Test Cases from Process Runs

1. From the left navigation menu, go to Processes.

2. Select a process.

3. Access a process run.

   1. To view a previous run, click on View Runs

Overview

In Kognitos, a process can call and run multiple subprocesses in parallel using the start a run syntax. Starting parallel runs enables the parent process to explicitly wait for the runs, retrieve the statuses, and fetch the outputs.

How To Start Parallel Runs

This section describes how to start parallel runs using the start a run syntax.

Syntax

Parameters

1. item: Object to loop over.

2. subprocess: The name of the subprocess.

3. input: Objects to pass to the subprocess, if necessary.

Example

Click to walk through an end-to-end example.

Handling Subprocess Runs

A parent process can explicitly wait for the runs of a subprocess to complete. Once all runs have finished executing, the system notifies the parent process which runs will proceed further. This allows the parent to respond accordingly, such as collecting data and output from each subprocess, or taking further action based on the results.

Adding an Explicit Wait

To add an explicit wait, include the following line into your automation after starting parallel runs:

Getting Output From Parallel Runs

To get results from subprocesses that are called with the start a run syntax, use the following syntax after adding an explicit wait:

Syntax

Below is a line-by-line overview of the automation syntax. Expand each line to learn more.

Examples

1. Singular Run

2. Multiple Runs

Walkthrough: Run Parallelization

This is an example walkthrough of invoking multiple instances of a subprocess in parallel runs.

Parent Process

Subprocess

Subprocess Output

In line 9, the greetings is referenced as output from the subprocess runs. The results are:

• Hello Eva

• Hello Ian

• Hello Leo

• Hello Nia

Overview

The Performance Dashboard provides a centralized view of your automation performance. It also calculates and displays your return on investment (ROI), giving you a clear picture of the value your automations are generating. To access the dashboard, click on Performance in the top navigation bar.

Daily Updates

The dashboard is updated daily to ensure you have access to the latest metrics.

(Updates occur every day at midnight UTC.)

Filters

The dashboard can be filtered by agents, published processes, or date ranges.

🗃️ Agent Filter

The agent filter allows you to narrow metrics by specific agents. By default, it includes all agents.

⚙️ Process Filter

The process filter allows you to narrow metrics by specific processes. By default, it includes all processes from all agents.

📅 Date Range Filter

The date range filter allows you to narrow metrics by time period. Filter options include:

• 6M: Last 6 months, including the current month

• 12M: Last 12 months, including the current month (default)

• YTD: Year-to-Date, including the current date

• Custom range: A custom range within a 2-year period

Dashboard Components

1. Value Insights

Value Insights offers an overview of the time and money saved by using Kognitos. These insights are calculated using the agents and processes selected in the . The summation (Σ) is taken across all selected processes using the date ranges selected.

Money Saved

A calculation of your dollar savings based on cost estimates configured for your processes.

Money Saved ($) = Σ (Total Completed Published Runs × Average Money Saved per Run)

Time Saved

A calculation of your time savings based on the time estimates configured for your processes.

Time Saved (Hours) = Σ ((Total Completed Published Runs × Average Time Saved per Run) ÷ 60)

2. Usage

Usage shows the total number of published process runs (regardless of completion status). The number is calculated based on the agents and date ranges selected in the . You can view usage Over Time or By Process by choosing the corresponding filter.

3. Completion

Completion metrics track the success of your published process runs. They are calculated using the agents and date ranges selected in the .

Completion Rate

The completion rate is the percentage of published process runs that have successfully completed out of the total number of published runs. It accounts for:

• Runs that completed automatically (STP)

• Runs that completed with user intervention (Guidance Provided)

Completion Rate (%) = (Total Completed Published Runs ÷ Total Published Runs) × 100

Efficiency

This metric calculates the Straight Through Processing (STP) rate, which is a measure of the runs that completed without user intervention. The STP rate is translated to an efficiency score, which represents how much more work can be completed with the same resources due to automation.

Efficiency Score = 100 ÷ (100 - STP)

Example

• An 80% STP results in 5x efficiency.

• With 80% STP, only 20% of the work remains manual.

• The same resources can handle 5 times more work.

4. Contract Details

Contract details provide an overview of your current contract, including:

📅 Key Dates

• Start date: Date your contract began

• End date: Date your contract ends

• Expiration: Remaining days until your contract expires (e.g., 35 days)

📊 Usage Summary

• Runs Used: Total number of published and test runs used (e.g., 151)

• Runs Remaining: Total number of published and test runs still available (e.g., 349)

• Total Allowed Runs: Maximum number of runs permitted under your contract (e.g., 500)

Downloading and Exporting Data

Kognitos makes it easy to share insights from your dashboard, without the hassle of screenshots or manual copy-and-pasting.

Save Images

You can download a PNG image of any individual metric directly from the dashboard:

1. Hover over any section (Value Insights, Usage, or Completion).

2. A download icon labeled "Download this section as an image" will appear in the top-right corner.

3. Click on the icon to download a .png of that section to your device.

Export Data

Click on Export Data to export an Excel report of your dashboard data. Kognitos will automatically email the .xlsx file to you. This feature is great for in-depth analysis, reporting, or sharing insights in spreadsheet format.

Configuring Your Dashboard for Value Insights

are generated off user-submitted cost and time estimates. While it's not required to provide estimates for every process, these metrics are only generated when the data is configured.

Configuring Process Estimates

To configure process estimates, follow these steps:

1. Click on Configure in the top right corner. This will open up a menu showing all your processes across agents.

2. For a given process, enter in the following details:

   • Time: Average time saved per process run (in minutes).

   • Cost: Average cost savings per run (in dollars).

1. Click Save to apply your configuration.

Setting Quarterly Goals

You can set quarterly goals for Value Insight metrics. To do this:

1. Click on Set Goal underneath either Money Saved or Time Saved.

2. Enter your goal (dollars or hours).

3. Click Save.

Overview

Kognitos simplifies email automation. With just a few lines of natural language syntax, you can send emails directly within your automated workflows — no configuration or server management is required.

Behavior