Interactive playground

See what your agent knows.

Explore how Restrct combines official knowledge and team memory into deterministic, source-labeled context.

03 / Context engine

Context assembled

→

Run a question to assemble source-labeled context.

04 / Agent

Agent

Waiting
◌

The agent receives controlled context after the evidence is checked.

Choose a path

Try a scenario

The architecture in motion

One playground. Six capabilities.

01

Knowledge

Turn official documents into operational facts.

02

Memory

Remember facts, events and relationships over time.

03

Context

Merge knowledge and memory into one source-labeled result.

04

Conflict Detection

Surface disagreements instead of hiding them.

05

Human Approval

Flag decisions that require human authorization.

06

Auditability

Preserve provenance from source to agent output.

A reproducible foundation

Same input. Same context.

Deterministic retrieval makes every agent decision inspectable and repeatable.