Our senior partners’ experience
Personal information detection.
Built into the review workflow.
An AI check built into the assessment platform gives reviewers a clear starting point for identifying personal information in supporting statements.
Our senior partners bring hands-on experience of building AI into existing systems. This integration is one example of their work before Ardento.
The challenge
Supporting statements describe a person's work and experience. They can also contain identifying details: names, contact information, locations and references to people or projects. Reviewers need to identify these details and decide what to remove before the assessment moves forward.
How the solution works
- 01
Read the supporting statement
When enabled, the AI check runs as the assessment team views a supporting statement. The results appear in the same review screen.
- 02
Surface potential identifiers
The AI checks for names, phone numbers, email addresses, locations, approver names and project references. Potential identifiers are grouped by type beside the original statement for reviewers to inspect.
- 03
Keep the reviewer in control
The reviewer checks the findings and decides what to redact through the existing workflow. When the AI reports no personal information, the interface still asks for a manual check.
Illustrative example
What a reviewer might see
Fictional text and sample findings, shown to explain the capability. This is not a product screenshot or a live AI result.
Supporting statement
I worked with Alex Morgan to improve our reporting process. We introduced a weekly review that helped the team identify missing information. For more details, contact me at alex.morgan@example.com.
Highlights connect the sample text to the findings shown here.
Potential identifiers
- Name
- Alex Morgan
- Email address
- alex.morgan@example.com
Next: the reviewer checks each finding and decides what to redact.
AI processing within AWS
The platform already uses AWS for its application infrastructure. The integration uses Claude through Amazon Bedrock, so statement analysis runs on AWS-managed infrastructure and the findings return directly to the platform's review interface.
Data protection and control
Model provider isolation
AWS hosts the model in accounts operated by Amazon Bedrock. AWS documents that model providers cannot access those accounts, customer prompts or model responses.
Encrypted requests
Statement text and findings travel over an encrypted connection to Amazon Bedrock, which requires TLS for communication.
Controlled access
Detection is restricted to assessment team members through a server-side role check. The platform also provides a setting to enable the capability.
Service protections: Amazon Bedrock data protection documentation.
Applied to your business
Where could AI help your team review information?
This is the practical experience we bring to helping your organisation integrate AI into its existing systems. We can help you find a useful starting point and build a solution with the right checks and human oversight.
Discuss your workflow →