Q&A: Paul Farrington of Glasswall, AI and Software Development Awards finalist

'The industry has been flooded with AI, but much of it prioritises speed over substance.'

The AI and Software Development Awards showcase outstanding achievements and solutions in the fields of AI and software development from organisations, personalities and teams.

This year's winners will be announced at a live awards ceremony on Thursday, 14^th May in London.

Foresight, Glasswall, is a finalist in Best Use of AI in Security category.

We caught up with Paul Farrington, CPO & CMO, Glasswall, to learn more about the company's work this year, which propelled it to the finals of the AI and Software Development Awards.

Paul Farrington leads the development and delivery of the company's cutting-edge cybersecurity solutions. With an outstanding track record spanning two decades in the SaaS software industry, he is the driving force behind Glasswall's innovative product portfolio. He is renowned for launching successful products and leading global sales-engineering teams.

Why do events like the AI & Software Development Awards matter?

They validate real-world impact. In a market full of AI claims, credible recognition cuts through and highlights who is actually delivering outcomes. For us, it's about showing that deterministic protection and explainable AI can set a higher bar for security.

What would winning this award mean to you, your company, your team?

For our team, it would be significant. We're not a large organisation, but we've taken on complex problems and built something genuinely different. Bringing Glasswall Foresight to market required technical ingenuity, persistence, and a clear focus on customer problems.

It would validate a long-term bet - starting years before generative AI became mainstream - that machine learning could be applied in a more rigorous, trustworthy way.

Most importantly, it would reinforce to customers that we are solving a critical challenge: identifying and neutralising malicious content with confidence and clarity.

What is your company's proudest achievement over the past year?

Our proudest achievement has been bringing Glasswall Foresight to market-shifting file security from reactive detection to predictive threat intelligence at the point of file ingress.

By combining patented CDR with proprietary AI trained on structural telemetry, we enable customers not only to neutralise threats, but to determine whether a file was weaponised before execution. The result is high-confidence prediction of unknown threats with low false positives - without relying on signatures or detonation.

What have been the biggest challenges of the year so far, and how have you overcome them? How have your people helped with that?

The core challenge has been making AI trustworthy in real-world environments - explainable, auditable, and deployable offline.

We addressed this by building proprietary models with transparent decision - making and proven provenance, validated against real-world conditions and designed to run without external dependencies.

Our people made the difference. Combining deep file-format expertise with AI engineering - and rigorously red-teaming our own models - they ensured we delivered something customers can rely on, not just something that works in theory.

How has your industry changed over the past year, and what changes do you think it still needs to make?

The industry has been flooded with AI, but much of it prioritises speed over substance. "AI-enabled detection" is often still reactive and opaque.

Customers are pushing back. They want solutions that are explainable, auditable, and reduce operational burden.

The industry needs to move beyond black-box models and retrospective detection. Security decisions must be transparent and made before threats execute - not after.

What do you see as the main opportunities for your industry in the coming year? How do you plan to capitalise on them?

The biggest opportunity is shifting security earlier - embedding it at the point where risk is introduced, especially as file-based threats scale with AI.

There's also a shift toward autonomous workflows, where ensuring AI systems interact only with trusted content becomes critical.

We're capitalising by evolving CDR into a more intelligent capability - combining deterministic protection with predictive insight. That means understanding intent before execution and embedding this into customer workflows.

In parallel, we're using agentic AI internally to accelerate development and continuously red-team our products - delivering faster innovation without compromising trust.

What are the key demands you have seen from your customers in the last 12 months?

Customers are demanding trust, control, and efficiency.

They want to understand why decisions are made, ensure alignment with their environment, and reduce noise. High false positives are no longer acceptable.

There's a clear shift toward solutions that combine protection with insight - making files safe while providing intelligence on whether they were part of an attack.

Which new technology trend are you placing your bets on?

We're betting on explainable and agentic AI - applied with discipline.

Explainability is essential where decisions must be defensible. Black-box models don't hold up in high-risk environments.

At the same time, agentic AI is reshaping how software is built. We're embedding it into our SDLC to accelerate development, automate testing, and continuously red-team our products.

The advantage isn't just smarter models - it's more trustworthy systems that improve over time.

The AI and Software Development Awards will take place on 14th May in London. Click here to view the shortlist and here to book your table.