Deterministic AI
A logic engine applies codified legal rules to structured facts and produces the same outcome for the same inputs, every time, with a complete reasoning trail. Neota is the logic engine behind governed legal and compliance solutions at global law firms and enterprise legal teams.
Definition
A legal logic engine is deterministic software that executes rules written by lawyers: statutory tests, eligibility thresholds, policy criteria, decision trees. Unlike a large language model, it does not predict a likely answer. It applies the rule. Same facts in, same determination out, with every step traceable to the specific rule that produced it.
Three properties define a logic engine:
No statistical variability. Outcomes are reproducible by design.
Every determination carries a reasoning record from input to conclusion.
Rules are authored, validated, and versioned by qualified professionals before anything executes.
Comparison
| Logic engine | LLM | |
|---|---|---|
| How it answers | Applies pre-validated rules | Predicts statistically likely text |
| Consistency | Same input, same output | Output can vary per run |
| Hallucination | Structurally impossible on the rules path | Inherent risk |
| Explainability | Full reasoning trail per outcome | Limited, post-hoc |
| Best at | Determinations, eligibility, compliance | Language, drafting, summarization, intake |
These are complements, not competitors. The failure mode is using an LLM to do the logic engine's job.
Process
Lawyers and compliance experts translate rules into structured logic using a no-code builder. No engineers required.
Logic is tested against known scenarios before it touches a real matter.
The engine applies rules to each set of facts, at scale, identically every time.
Every outcome ships with an audit trail: inputs, rules fired, determination reached.
Matters outside defined parameters route to a human, with the analysis attached.
Architecture
Modern legal AI architecture splits the work. Generative AI handles the conversation: intake, drafting, explanation. The logic engine handles the determination: whether the threshold is met, which obligation applies, what the outcome is. LLMs operate inside the structure; they do not drive outcomes. This is the architecture behind regulator-approved technology-only law firms, and it is how Neota deployments run today.
Interface layer
Generative AI
Intake, drafting, explanation
Determination layer
Logic engine
Thresholds, obligations, outcomes
Record layer
Audit trail
Inputs, rules fired, determination
Use cases
Why now
In May 2025, the SRA authorized Garfield.Law, the first technology-only law firm in England and Wales. In February 2026, it authorized LawFairy, the first built on a fully deterministic legal model: regulated outcomes produced by pre-validated rules with a complete, auditable reasoning trail, not probabilistic AI. The regulatory signal is that software can deliver legal outcomes when those outcomes are rule-based, consistent, and auditable. That is the logic engine model.
FAQ
It is a form of AI, sometimes called symbolic or deterministic AI. It predates LLMs and behaves differently: it reasons from rules rather than predicting from patterns.
No. On the pure rules path there is nothing to hallucinate; outcomes come only from validated rules applied to provided facts.
On Neota, no. Logic is authored in a no-code environment by legal and compliance professionals.
Yes. The standard pattern is LLM for language and interface, logic engine for determination, with your own models and API keys.
Bring one process: an eligibility test, an intake policy, a compliance check. We'll show it running deterministically, audit trail included.