AI product case study

Legal research that shows its sources.

How Quindart built Jurist.ma as an AI-assisted legal research and document-analysis platform centred on Moroccan legal context and verifiable sources.

Start with the Readiness Audit

The challenge

A fluent legal answer is useless if the evidence cannot be checked.

Product decision

The product should never create the appearance of certainty when its evidence is incomplete. Source-supported information and clear limitations are part of the user experience, not a technical detail hidden behind it.

Before

How the work moved.

  • General-purpose answers could sound certain without a controlled Moroccan legal source base.
  • Users had no dependable route from an answer back to material they could inspect.
  • Private document analysis needed a boundary from public legal knowledge.

After

The route now in use.

  1. A user asks a question or uploads a document for review
  2. The platform finds relevant material from the appropriate context
  3. The response is generated with that context and available sources are surfaced
  4. The user reviews the information and remains responsible for professional judgment

A legal question needs more than a polished reply.

Jurist.ma was conceived as an AI-assisted legal research and document-analysis platform focused on Moroccan law. The aim was not to make a chatbot sound convincing. It was to help lawyers, companies, entrepreneurs, students, and citizens reach relevant material with enough context to review it properly.

That distinction matters. General AI systems can answer confidently without a controlled knowledge base, reliable sources, or a clear boundary between information and formal advice. In legal work, those gaps can make an answer less useful precisely when it sounds most certain.

Retrieval and interface design had to work together.

The product needed to organise legal material so the right context could be found before a response was generated. It also needed to present that context in a way that did not overwhelm the person asking the question.

Quindart designed the research flow around a simple sequence: ask a question, retrieve relevant material, produce a structured response, and show the sources that are available for review. The interface supports exploration, while the source trail gives the user a practical way to test what they are reading.

Document analysis needed a separate trust boundary.

When a user uploads a document, the platform processes it so they can ask about clauses, sections, and meaning in context. That private document context remains separate from the public legal knowledge base.

This design recognises that uploaded legal material may be sensitive. It also makes the product's role clearer: it helps users inspect and understand information, but it does not replace the professional review or advice that a legal matter may require.

The result

Trust built on sources you can check.

Faster access to relevant legal information and topics
A more transparent way to explore answers than a general purpose chatbot provides
A useful first pass for long documents without treating the result as legal advice
A searchable product experience built around an existing legal content library

Capability proof

What this demonstrates

The reusable operating strengths behind the visible product.

Source-linked answers
Specialist document retrieval
Evidence verification
Reviewable AI output

Your next step

Start with the Readiness Audit

Assess the structured information, permissions, outputs, and verification an agent-facing capability requires.

Start with the Readiness Audit