Test Updates Safely: CyDeploy’s System Replica at Disrupt 2025
▼ Summary
– Companies face a challenge in balancing quick security patching with ensuring system updates don’t disrupt operations.
– CyDeploy uses machine learning to create a “digital twin” replica of company systems for safe update testing before deployment.
– The system records and labels daily application usage, with human administrators verifying the machine learning interpretations for accuracy.
– CyDeploy helps write test scripts faster and focuses on critical servers and Tier 1 applications requiring careful security updates.
– Customers can choose between CyDeploy’s internal large language model or OpenAI’s model, with options to keep proprietary data on their servers.
For businesses relying on software, maintaining security without disrupting operations presents a constant challenge. CyDeploy offers a solution by creating a digital twin of critical systems, allowing companies to test updates safely before implementation. This approach helps organizations patch vulnerabilities rapidly while avoiding potential system failures that could impact daily business functions.
Founder Tina Williams-Koroma developed CyDeploy to address this critical balancing act. The platform employs machine learning to observe and document how company systems operate during normal use. By recording user interactions and application behaviors, the technology builds an accurate replica environment where updates can undergo thorough testing.
The system captures screen activity and applies automated labeling through its machine learning capabilities. However, CyDeploy maintains human oversight throughout this process. System administrators with deep knowledge of their organization’s operations verify the machine learning interpretations, ensuring accuracy and preventing potential misinterpretations.
Williams-Koroma emphasized that this methodology significantly accelerates test script creation. Administrators can deploy updates into the digital twin environment and execute functional tests that mirror real-world usage patterns. This enables them to identify potential issues before any changes affect production systems.
Companies implementing CyDeploy have flexibility in how they utilize the technology’s analytical capabilities. Organizations can choose to run CyDeploy’s proprietary large language model within their own infrastructure, keeping sensitive data contained. Alternatively, they may opt to use OpenAI’s model, though this would involve transmitting information outside company networks.
The platform primarily targets critical infrastructure rather than individual workstations. Williams-Koroma noted that CyDeploy proves most valuable for what she termed “Tier 1 applications”, systems requiring prompt security updates but where operational stability remains equally important. These typically include servers and essential machines where changes carry significant business consequences.
CyDeploy’s innovative approach has earned the startup recognition as a Top 20 finalist in TechCrunch’s Startup Battlefield competition. The company will showcase its technology alongside numerous other emerging companies at the upcoming Disrupt conference, providing attendees opportunity to evaluate this novel solution for managing system updates securely.
(Source: TechCrunch)
