Want something like this?
Tell us what yours would need to do differently. We reply with a scope and a price.
Work · 05 · Explainable AI
Mizaan
A multi-agent scoring engine built on one non-negotiable rule: every score the system produces must be explainable to a human.
Status
In development
In active development; methodology published
Domain
AI evaluation and scoring
The problem
Organisations increasingly let AI systems score things that matter: sales calls, support interactions, student work, engineering output. Most of these systems are black boxes. A number comes out and nobody can say why. If you cannot explain a score you cannot appeal it, audit it or improve it. You can only obey it.
The approach: three layers
Mizaan is Arabic and Urdu for balance, the scale of justice. No single model’s judgement is ever the final word.
- Primary scorer: evaluates the input against explicit, human-readable criteria and returns a score with its reasoning attached.
- Auditor: an independent agent challenges inconsistencies and weak justifications before anything is reported.
- Human validation: disputed or low-confidence scores go to a person, and corrections feed back as reference material for future cases.
The system’s judgement converges toward its human overseers’, visibly, case by case, instead of through opaque retraining.
Published
The methodology is a sole-authored preprint: When Disagreement Means Learning (or Bias): Human-Governed Memory Consolidation and Counterfactual Diagnosis in Multi-Agent LLM Scoring, Zenodo, June 2026, CC-BY-4.0, with an open evaluation harness anyone can run.
Status
In active development, evaluated across two contrasting domains over three independent runs for the published study. Current focus: hardening the auditor layer and expanding the criteria library.