PRIVILEGE AI / RESEARCH & ENGINEERING
Intelligence,
under scrutiny.
We develop specialised models and study how they reason,
retain knowledge, and evaluate the work of other models.
SCROLL TO EXPLORE ↓MODELS · KNOWLEDGE · SYSTEMS
SMALL LOCAL MODELS
16 OCT 2026Small models.
Grounded in the source.
Small local models, each specialised for one task. They run offline on ordinary hardware, and every output carries the exact quote from the source it came from.
FORMAT-EXACT ANSWERS
Exact matches, out of 24.
- Small model, untuned
- Larger general model
- MKMM
Results in full +
| Task | Small model, untuned | Larger general model | MKMM | Challenge set |
|---|---|---|---|---|
| Fact extraction | 0/24 | 12/24 | 18/24 | 24/24 |
| Authority check | 3/24 | 7/24 | 24/24 | 24/24 |
| Consistency check | 7/24 | 13/24 | 24/24 | 24/24 |
01 / MODEL RESEARCH
Developed, fine-tuned and evaluated in-house.AI that questions
AI.
MKC2 is a specialised supervision model developed by Privilege AI: a model that evaluates AI-generated work and the actions of AI agents. It runs entirely offline.
MKC2 is the first fully offline, fine-tuned supervision model built specifically to judge legal drafting and legal-agent actions. It ranks defective work above sound work at 0.962, against 0.728 for the base model it was fine-tuned from.
Explore the measured results ↓MEASURED. NOT ASSUMED.
It finds the flawed work.
96/100
Put one flawed passage next to one sound passage. MKC2 picks out the flawed one 96 times out of 100.
The stock model it was built from manages 73. The model it replaces, 65.
Taken on its own
MKC2 reports and flags.
It does not block or act.
Checks we refuse to rely on
One check could not be shown to work,
so it is switched off, not shipped.
Held-out items
None of them seen during training.
45 separate checks on each.
Calibration record, 21 September 2026. Every figure is measured on held-out items, with the model it replaces re-scored in the same run on the same material. No attorney dispositions have been recorded, so none of this is evidence that MKC2 agrees with a lawyer on live work. MKC2 reports; it does not act.
Results in context +
MKC2 leads on every measure in the record against both the model it replaces and the stock base model, including how far its stated confidence can be trusted.
The gain is largest on the measure that decides whether a confidence figure is usable at all: how well the model ranks defective work above sound work. The model it replaces scored 0.6457 there, below the stock base model.
Of 45 questions, one is refused outright for insufficient ranking quality and 25 carry report-only authority, changing nothing. Questions that cannot be shown to work are not relied upon.
MKC2 runs in review-only mode: it reports and flags, it does not block or act on its own. That boundary is deliberate. These are benchmark results on held-out data, not live-matter outcomes.
02 / RESEARCH DIRECTIONS
Questions we are working on.MODEL SUPERVISION
When should a model
trust another model?
We study the reliability of AI-generated work and the role of specialised models in reviewing it. Our work in HIGHCOURT brings these questions into a demanding application.
Read the evaluation note ↗EVALUATION & AUTHORITY
Capability is measured.
Authority is earned.
Strong benchmark performance is one part of responsible deployment. We distinguish measured capability from the responsibilities a model should be given.
The reported model operates in review-only mode.
03 / RESEARCH & ENGINEERING
The work spans the full system.Models.
Knowledge.
Infrastructure.
We work across model adaptation, retrieval and private deployment. Each layer changes what a system can do, what it can know, and how its behaviour can be evaluated.
Training and
fine-tuning.
Dataset construction, model training and fine-tuning, with task-specific evaluation and held-out comparisons.
Retrieval and
knowledge.
Retrieval-augmented generation and knowledge systems. Research into the relationship between context and model performance.
Inference and
systems.
Local inference, application integration and on-premise infrastructure. Security boundaries considered throughout the system.
04 / PRIVATE SYSTEMS
Control is part of the architecture.Private inference.
Explicit boundaries.
Deployment is part of the research problem. Data access, model authority and operational constraints shape how AI behaves outside a benchmark.
On-premise inference.
Local model execution with explicit hardware, network and operational constraints.
Access with intention.
Permissions and retrieval boundaries define which information a model can access and which actions it can propose.
Security from the start.
Sensitive data, model behaviour and deployment risk are considered throughout development and evaluation.
CONTROLLED ENVIRONMENT
Data — Models — Applications
ONE CONTROLLED BOUNDARYPRIVATE SYSTEMS
17 OCT 2026Intelligence on top of retrieval.
Not an LLM wrapping a SERP.
Private-beta multi-source search. A proprietary search-intelligence layer sits on top of retrieval. Search keeps working even when the overview model is unavailable. Not a full independent web index.
THREE SPEEDS
- Instant — people, places, brands and quick lookups: a structured answer card, returned straight away.
- Intelligent — comparisons, research, high-stakes questions: a grounded overview, cited only to pages actually retrieved.
- Deep — multi-part or conflicting questions: a more thorough search, then the same grounded overview. Bounded, not an open agent.
WHAT WE MEASURED
Ranking is measured on a labeled held-out set (17 Oct 2026), not assumed. Changes that did not clearly improve results were not made the default. Live ranking quality is not claimed. Grounded overviews return in seconds; simple lookups return instantly.
BEHAVIOR
Official sources are surfaced first where they exist. Overviews that cannot be backed by the retrieved pages are not shown.
PRIVATE SYSTEMS
21 OCT 2026Shared matter state.
Specialist systems, connected.
LMMES (Legal Matter Modular Execution System) is now integrated inside HIGHCOURT. It connects specialist AI, structured legal knowledge, and versioned matter state, running locally and offline.
It brings together the systems we have built so they share a persistent matter state:
- Source analysis that keeps every finding tied to its source
- Independent critique of generated work
- Structured legal review and assessment within HIGHCOURT’s analysis workflow
- Legal ontology and retrieval connected to HIGHCOURT’s existing legal resources
- Evidence and work product organised around each matter
Matter state is versioned, and every step stays reviewable.
Acceptance is attorney-controlled. The system proposes; the attorney decides.
The integration operates through HIGHCOURT’s existing Add Matter → ingestion → Run Analysis workflow and its ongoing review processes.
Local inference. Local retrieval. Encrypted matter storage. Versioned matter state. Attorney-controlled decisions.
Learn more about HIGHCOURT →05 / ABOUT PRIVILEGE AI
Experience compounds.OUR FOUNDATION
Built on decades
of experience.
A continuous practice
of engineering.
Built on decades of experience in software engineering and data science, Privilege AI brings that foundation to the development of specialised models and AI systems.
We connect that experience to the whole system: how data is structured, how software is built, and how intelligence becomes useful.
Web developmentTechnical SEOData scienceAI engineering
CONTACT
Start a
conversation.
For research, technical enquiries and collaborations.
Tell us a little about what you’re working on.