← All projects
Juris CRM / AI + workflow

From first contact to legal operations.

Contextual proposals, human review, approval, and documents connected to the CRM.

Explore the full demo ↗
Next.jsTypeScriptIA contextualPDF
● ● ● Juris CRM · synthetic demonstration
Juris CRM demo screenshot
01 / THE PROBLEM

What needed to work

Preparing proposals and converting a lead into actionable work requires consistent context across sales and operations.

02 / CONTRIBUTION

My contribution

CRM, proposal, document, and operations interfaces assembled into an anonymized demo. Source contribution and portfolio adaptations are recorded in the source inventory.

03 / ARCHITECTURE

How the pieces connect

  1. 01Full CRM interface
  2. 02Original API contracts
  3. 03Isolated PostgreSQL state
  4. 04Validated inference + approval
  5. 05PDF from current state
04 / DECISIONS

Engineering choices

  • Explicit approval before the final document.
  • One repository serves server pages and APIs.
  • Legal data, companies, and amounts are entirely synthetic.
05 / DEMO SCRIPT

Two ways to explore

1Lead → proposal → review → approval → PDF.

2Conversion → customer → work order → tasks → Kanban.

06 / SOURCE AND ADAPTATION

What the demo demonstrates

Original: Legal CRM, proposals, tasks, and operations. Original client identity and data were removed.

Demo: Fixtures, visitor isolation, reset, translation, demo generation, and inference adapter.

GitHub · source repositories may be private ↗
07 / LIMITATIONS

Clear boundaries

The demo provides no legal advice. Simulation is the default; live inference requires an authorized session. Integrations and external acceptance are simulated.

State belongs to your visitor session. Restart from the demo bar.

See it working.

Open demo ↗