01Internal pilot
Shadow Judge
Structured AI adjudication and actionable feedback for debate, with a human judge review process.
Independent thinking. Applied AI.
Better arguments.
Better decisions.
AI should help us think more clearly.
Not do our thinking for us.
We build tools that bring structure to ideas, substance to feedback, and human judgment into the loop.
Built in Canada.
Made for a world of perspectives.
The counterpoint
Two perspectives.
A stronger understanding.
Our starting point
Good judgment begins with a well-examined argument.
01 · Our work
One foundation.
Three directions.
Beginning in debate education.
Exploring what better reasoning can make possible beyond it.
02In exploration
Public education
A path to make thoughtful debate practice and reasoning feedback accessible to more learners.
03In exploration
Syllogy Enterprise
Applying structured challenge and evidence-linked evaluation to complex business decisions.
02 · Where we begin
Built around
real debate.
A working platform.
A real learning environment.
Shadow Judge is being tested internally at Vancouver Debate Academy, through individual practice and an in-house tournament.
Our focus is practical: useful ballots, a dependable human judge queue, clear annotations, and a platform that works when it matters.
Vancouver Debate AcademyInternal testing environment
The review loopShadow Judge
- 01
Make the argument.
Debaters put their reasoning into practice within a defined debate format.
- 02
Examine the ballot.
AI generates structured evaluation and feedback against the selected rubric.
- 03
Bring in human judgment.
The human judge queue supports review, annotation and correction.
- 04
Learn. Refine. Repeat.
Testing reveals where feedback, reliability and the review process need to improve.
Evaluation is a process, not just a score.
03 · Our approach
More than an answer.
A way to examine it.
Our work starts with a simple belief: the reasoning behind a result matters as much as the result itself.
Structure before scoring.
Different contexts call for different rules. Evaluation should reflect the selected format and its explicit criteria.
Feedback you can question.
Useful feedback invites examination. We are testing how human review and annotation make that possible in practice.
Evidence before promises.
We test in real settings, document limitations and improve deliberately. Internal testing is a beginning—not a claim of universal validation.
04 · A Canadian direction
Build here.
Keep control here.
Canadian sovereignty is a development objective—not a label we apply lightly.
Our long-term aim is Canadian-controlled processing and storage across the full pipeline: audio, transcription, evaluation, generated data and backups.
We are working toward a deployment model that institutions can assess on its evidence, not simply its address.
Our intended direction
01
Processing in Canada
Transcription and inference running on Canadian-hosted infrastructure.
02
Control across the data lifecycle
Clear boundaries for access, storage, retention and deletion.
03
Accountability that can be checked
Documented dependencies, operational controls and independent review.
These are roadmap objectives. They are not current certifications or guarantees of compliance.
05 · The next conversation
Progress starts
with a counterpoint.
We are interested in conversations with educators, institutions and partners who want to explore what thoughtful AI can make possible.
info@syllogylabs.comA short introduction you can share with your network.