How Can AI Agents Adapt to Real-World KYB Operations? Six Months of ARGOS Omni ②
In Part 1, we explored how ARGOS Omni has evolved over the past six months, focusing on the features that help teams understand, trust, and verify AI-generated review results.
From PDF reports and field-level explanations to Analysis Score, which helps teams prioritize reviews, we looked at how Omni makes AI-driven decisions more transparent and verifiable.
But being able to verify AI-generated results doesn't solve every operational challenge.
In real-world KYB operations, every organization has different review policies, verification requirements, and workflows. Some businesses may need to prioritize business registration checks, while others may require bank account ownership verification or additional document reviews.
To support these different requirements, AI Agents need to do more than produce results. They must also be configurable and manageable within an organization's existing operations.
In Part 2, we'll explore the latest updates designed to make ARGOS Omni more flexible and adaptable to real-world KYB workflows.
From workflow customization and expanded verification capabilities for Korean businesses to credit usage tracking and audit records, here's how Omni has evolved over the past six months.
Customize Workflows to Match Your Business Requirements
ARGOS Omni allows users to create review workflows by describing their policies in natural language.
This makes it possible to build automated review processes without complex development work. However, once a workflow is deployed, businesses may need to adjust specific steps as their operational requirements evolve.
For example, a particular step may be assigned a verification engine that doesn't fully match the intended policy, or a new requirement may call for additional verification.
Previously, making these changes could require rewriting the policy and regenerating the workflow.
However, modifying an entire policy can also affect steps that were already working correctly, creating additional work to review and validate the updated workflow.
Edit Only What Needs to Change
Omni now allows users to modify the verification engines assigned to individual workflow steps.
By selecting the n Engines button next to an action, users can browse the full engine catalog and choose the engines that best match their requirements.
Multiple engines can also be assigned to a single action, with engine names and descriptions available to help users understand their functions.
For example, if a business needs to change the engine used to verify company information, it can update that specific step without rewriting the entire policy.
Users can also modify the instructions associated with each action through the Edit Notes feature.
In some cases, unexpected review results may not be caused by the policy itself, but by how the policy was translated into the workflow's action-level instructions.
By editing these notes directly, teams can make targeted adjustments without regenerating the entire playbook.
This means businesses can quickly generate workflows using natural language while maintaining direct control over individual configurations during ongoing operations.
Additionally, four new AI model options have been introduced during workflow creation, providing greater flexibility for different business requirements.
Expanded Verification Engines for Korean Businesses
In KYB operations, reviewing the information contained in submitted documents is only part of the verification process.
Organizations also need to confirm whether that information matches official records and remains valid.
For example, even if a business registration certificate contains a valid-looking registration number and representative's name, reviewers may still need to confirm whether the business is currently active.
Similarly, a submitted bank document may contain an account number and account holder's name, but additional verification may be required to confirm the actual account ownership.
Traditionally, these checks could require reviewers to access separate verification systems, retrieve the relevant information, and manually record the results.
Even when document analysis was automated, these external verification steps could introduce manual work back into the process.
From Document Analysis to Official Data Verification in a Single Workflow
To support these processes, Omni has introduced three verification engines specifically designed for Korean businesses.
Verification Engine | Key Function |
|---|---|
Business Registration Status Check - KOR | Checks business registration status, including active or closed status, and tax classification |
Business Info Verification - KOR | Verifies whether the business registration number, representative's name, and business opening date match official records |
Bank Account Owner Name Lookup - KOR | Retrieves the account holder's name using the bank code and account number |
When a policy includes requirements such as checking business registration status or verifying bank account ownership, Omni can automatically assign the relevant engines to the workflow.
For example, a workflow can extract a business registration number from a submitted certificate and use that number to check the company's current registration status.
Similarly, bank account verification can be incorporated into the same workflow to compare submitted information against the account holder details returned by the verification engine.
This enables organizations to connect document data extraction and external data verification within a single, continuous workflow.
Omni has also introduced an Address Validation engine to verify whether submitted mailing addresses exist and standardize address information.
Please note that the - KOR engines support Korean businesses and bank accounts only. Additionally, South Korea is not currently included among the 40 regions supported by Address Validation. Users should confirm each engine's geographic coverage when configuring their workflows.
Gain Greater Visibility into Credit Usage and Audit Records
Once an AI Agent is deployed in day-to-day operations, automation performance is only one part of the equation.
Organizations also need to understand operating costs and track what happens throughout each analysis.
This is particularly important for businesses managing multiple projects or workflows, where different verification processes may generate different levels of usage and cost.
To improve operational visibility, Omni has introduced a dedicated credit usage section within project management.
Understand Where Your Credits Are Being Used
The credit usage tab provides period-based usage summaries, transaction-level details, and CSV export functionality.
Each transaction includes the BASE unit price and quantity, allowing users to understand how the final credit amount was calculated.
For example, when an analysis involves multiple verification engines, the analysis execution itself is recorded as an Operation, while individual engine usage is recorded separately under Tools.
Rather than simply checking the total number of credits consumed during a month, organizations can identify which workflows and individual steps are contributing to their operating costs.
If a particular step repeatedly calls a high-cost engine, teams can review that workflow and explore ways to reduce unnecessary calls.
CSV exports can also support internal reporting and cost allocation across departments or projects.
Trace Each Analysis Through Detailed Audit Records
Alongside credit usage visibility, Omni has also enhanced its audit capabilities.
The Audit tab has been reorganized around summary information and step-level records, including external engine call details.
Teams can review which tools were called during an analysis, what inputs were provided, and what results were returned.
For example, if a specific business review requires further investigation, reviewers can go beyond the final result and trace the verification steps that were actually performed.
Token usage is also available at the task level through the Info tab and aggregated views in the Audit tab.
Together, these capabilities help organizations move beyond basic usage monitoring toward more structured management of AI Agent operations, costs, and verification records.
Making AI Agents More Adaptable, Manageable, and Accountable
In Part 1, we introduced the features that help people understand and verify the results produced by AI Agents.
In Part 2, we explored the next step: configuring and managing AI Agents to fit real-world business operations.
From editing individual workflow steps and expanding verification capabilities for Korean businesses to monitoring credit usage and reviewing audit records, Omni's development over the past six months has gone beyond expanding the scope of automation.
The focus has been on creating an environment where organizations can apply their own review policies, understand what happens during execution, and make adjustments whenever human intervention is needed.
While AI Agents handle repetitive document analysis and verification tasks, experienced reviewers can spend more time investigating exceptions and making decisions that require human judgment.
Ultimately, ARGOS Omni is not designed simply to remove people from the review process.
It's designed to let AI Agents handle repetitive reviews while enabling people to focus on higher-value decisions and more complex work.
Want to learn more about how ARGOS Omni works and how it can support your KYB operations?
