Senior User Experience Designer and Accessibility Expert
I'm a UX specialist, and I love improving systems and simplifying the complex.
Glasswall Foresight: UX strategy for an AI-powered threat prediction feature
Shortlisted for "Best Use of AI in Security" — AI & Software Development Awards 2026
Problem statement:
Glasswall Meteor's CDR technology neutralizes threats inside files and delivers safe, clean output. But it couldn't tell users whether the original file was malicious. Users had no way of knowing if they were under attack.
This gap was especially significant for files processed under permissive policies, files that couldn't be fully cleaned, and environments where understanding the threat landscape was operationally critical. Foresight was Glasswall's answer: a proprietary AI-powered threat prediction capability built on five-plus years of data science research.
My role:
Sole UX designer, end-to-end, from discovery through to shipped product. I joined at the start of the project, shaping the UX strategy before the feature had a defined form in the interface. Responsibilities included defining where and how Foresight would surface in Meteor, designing the threat feedback system, collaborating with engineering across sprint cycles, and ensuring designs met WCAG 2.2 AA requirements.
Users and goals:
Meteor's user base spans security IT administrators and SOC team members, everyday desktop users, and professionals in legal and financial services. Foresight had to work for all of them.
-
Security professionals and IT admins: understand the threat signal with precision and integrate it into Zero Trust workflows
-
Business and general users: receive clear, unambiguous guidance about a file without needing to understand the AI model behind it
-
Buyers and leadership: see evidence that Glasswall's AI is purpose-built and meaningfully reduces risk
Design strategy:
Foresight as a distinct capability - Because Foresight may eventually appear across other Glasswall products, it needed to feel self-contained rather than just an extension of CDR, with its own visual identity while still fitting within Meteor's established design patterns.
Three surfaces, all my placement decisions
-
Settings page: a toggle to enable or disable Foresight, making activation status immediately transparent to administrators
-
Processed files table: threat status surfaced inline with file activity, keeping Foresight in the core workflow rather than a secondary screen
-
File analysis report: the full Foresight output, presenting one of three result states — No Threats Detected, Suspicious, or Malicious — with policy-aware guidance and file availability controls tied to each
Communicating risk without revealing the model - Surfacing raw scores or numeric probabilities was ruled out to protect Glasswall's patented AI model from reverse engineering. Instead, a plain-language, three-state feedback system communicates confidence in human terms while keeping the model's internals opaque.
Making the role of AI visible - When Foresight was the reason a file received a particular status, users needed to know AI was involved. A visual indicator was developed to mark these occasions consistently across the interface.
Serving all users equally - The three-state system, combined with contextual guidance that adjusts based on policy settings, allows a security professional and a general user to both understand a result without the design being condescending to one or oversimplified to the other.
Design process:
The shipped design integrates Foresight across three surfaces with a consistent visual language distinguishing AI-driven threat intelligence from standard CDR output:
-
A settings toggle for administrators to control Foresight activation
-
Threat status surfaced inline in the processed files table
-
A file analysis report section with three differentiated result states and policy-aware guidance
-
A visual AI attribution indicator for occasions where Foresight determined the file's risk status
Impact & recognition:
-
Foresight shipped in January 2026 to positive reception from customers and internal stakeholders. Within a month of release, it was shortlisted for the AI & Software Development Awards 2026 in the category "Best Use of AI in Security."
-
The feature represents the first deployment of Glasswall's proprietary AI model in a product UI, establishing design foundations and patterns that will carry forward as Foresight expands to other products.
Reflection:
-
The iterative sprint-based process was the right approach for a feature this novel, allowing the design to improve alongside the AI model. The award shortlist, arriving so quickly after launch, is a strong signal that the work landed well.
-
One thing this project reinforced: getting information architecture right early, deciding where a feature surfaces before designing the feature itself, makes everything downstream easier. Those three placement decisions shaped the entire design system for Foresight.
