01 / Research
Automating the work
of investigation.
Our current work is in automated cybersecurity: systems that examine software for weaknesses and gather the evidence needed to decide whether those weaknesses are real.
Automated cybersecurity
Software leaves clues in its code, interfaces, and behaviour. Finding one is the beginning of an investigation. The harder work is establishing what it permits, under which conditions, and whether the result can be repeated.
We're developing automation around that process, with the aim of producing findings that a person can inspect and understand.
- ObserveWhat is the system actually doing?
- ChallengeWhich assumption changes the answer?
- ReproduceDoes the behaviour survive another attempt?
- ExplainWhat does the evidence establish, and where does it stop?
Inspect a small example
02
Questions within the question
Can an automated investigator disprove its own first finding?
What makes evidence transferable between different systems?
How much of an explanation can be checked without trusting its author?
03 / Publications
The public record
Research notes, methods, and results will be published here as they become ready to share.
If you're working on a related question, we'd like to compare notes.