Vibe Coded Proejct

Job ID: 40256173

Budget: €30 – €250 EUR

PROJECT TITLE

AI Legal Case Intelligence Platform (Ireland) – Phase 1 MVP (Multi-Tenant Demo + Pilot Build)

PROJECT SUMMARY

We need a Phase 1 MVP of an AI-powered Legal Case Intelligence Platform for Irish solicitor firms.

The platform is a mini legal CRM + case intelligence system:

Store and manage cases (matters)

Upload and organise evidence (emails, WhatsApp, PDFs, images)

Auto-generate chronology/timeline

Map evidence to legal factors (Ireland family law first)

Identify evidential gaps

Analyse communications patterns

Provide an AI query console to ask questions across the case data

This is Phase 1 only.
Phase 2/3/4 will be awarded to the same developer if Phase 1 is built correctly.

REQUIRED PRE-BID ACTION (NON-NEGOTIABLE)

Before bidding, the developer must:

Create a free account at: https://www.courtready.family/

Test the free version flow and features (upload, timeline, gap detection, simulator style).

Confirm in the bid that they tested it.

We are not copying it, but it’s a benchmark for the style of tool and expected MVP scope.

ARCHITECTURE DECISION (IMPORTANT)

Phase 1 must be built as a central hosted system (Option A) with multi-tenant separation per firm.

Reason: we need a demo + pilot platform where we can iterate and update quickly.

Future (Phase 2+): we may move paying firms to dedicated deployments (VM per firm).
So Phase 1 must be built in a way that can later support “dedicated instance per firm” without rewriting the platform.

Key requirement: strict tenant separation (firm-level data isolation).

PHASE 1 GOALS

Deliver a working MVP that:

Is usable for real solicitor work (pilot testing)

Stores cases and evidence permanently

Builds structured outputs from evidence

Allows natural language querying of each case’s dataset

Is designed for easy future upgrades and deployment flexibility

PHASE 1 FUNCTIONAL REQUIREMENTS
1) Firm Dashboard (Mini CRM)

A firm user logs in and sees:

Total matters

Active matters

Matters by status (e.g., Awaiting Documents / Drafting / Review / Ready)

Recently updated matters

Global search (search by client name, matter name, keyword)

Matter list rows show:

Matter name / Client name

Matter type

Status

Stage

Last updated

Assigned staff member (basic)

Matter fields (minimum):

Client name

Client contact details (phone/email)

Opposing party name

Matter type (family law for Phase 1)

Stage (dropdown)

Status (dropdown)

Internal notes

2) Matter Workspace (Per Case)

Each matter contains these tabs/modules:

A) Evidence Vault (Upload + Organise)

Uploads supported:

PDF

DOC/DOCX

Images (JPG/PNG) – OCR required

WhatsApp export (.txt)

Emails (paste text OR upload .eml)

For each upload, system must store:

Original file

Extracted text

Date(s) detected

Source type (WhatsApp / email / letter / report / photo)

Tags/categories (editable)

Minimum categories:

Access/Contact

Communication

School

Medical

Compliance

Other

B) Auto Chronology Builder (Core)

System must auto-create a chronology from extracted evidence:

Timeline table view (date + event + source reference)

Ability to edit events manually

Ability to filter timeline by category/tag

Export chronology (PDF or DOCX)

C) Legal Factor Mapping (Ireland – Family Law for Phase 1)

System maps evidence to a structured factor checklist relevant to Irish family law.

Outputs should show:

Each factor

Evidence items supporting it

Factors with weak/no evidence

Example factors (can be refined):

Child welfare / best interests

Stability and routine

Communication/co-parenting conduct

Access facilitation/interference

Any risk/concern indicators

Practical arrangements

D) Gap Detector (Core)

System must generate a structured “gaps” output:

Missing documents

Unsupported claims

Weak areas needing corroboration

Suggested next info to request (high level)

This must be presented as checklists/lists, not vague paragraphs.

E) Communication Pattern Scanner (WhatsApp/Email)

For message threads, output:

Response delay patterns

Refusal/no-counterproposal patterns

Escalation trends (tone shifts)

Summary of key recurring disputes (e.g., access scheduling)

3) AI Query Console (MANDATORY IN PHASE 1)

Within each matter, user can ask questions like:

“Summarise the access issues.”

“List evidence supporting missed contact claims.”

“What are the key gaps?”

“Summarise communications between Jan–Mar.”

“What is the latest update and status?”

This must use a retrieval approach (RAG/embeddings) so answers are grounded in stored matter data.

Important: the platform must store case data and allow intelligent retrieval from it.
We are not looking for generic ChatGPT answers.

4) Outputs (Generate Structured Case Pack)

Phase 1 outputs (minimum):

Chronology (table)

Evidence index (list of uploads with tags)

Gaps report

Legal factor map summary

Communication pattern summary

Export formats acceptable:

PDF + optional DOCX

DATA STORAGE REQUIREMENTS (PHASE 1)

Platform must permanently store:

Matters

Evidence files

Extracted text

Metadata

Timeline events

Analysis outputs

Embeddings / vector index per tenant

Multi-tenant separation required:

Every record must be tied to firm/tenant ID

Tenant data must not be visible across firms

DEPLOYMENT & UPDATE REQUIREMENTS

Phase 1 is central hosted (multi-tenant).
We need the ability to update quickly.

Strong preference: Dockerised deployment so we can later run:

Central hosted platform (now)

Dedicated instance per firm (future VM option)

Developer must provide:

Source code

Docker setup (or clear deployment steps)

Basic admin instructions to create firms/users

Notes on how updates will be deployed

But Phase 1 must be built so these can be added later.

FUTURE PHASES (FOR CONTEXT ONLY)

Phase 2:

Task system + reminders + “awaiting documents” automation

Courtroom Simulator style module (cross-exam prompts based on case evidence)

Cross-case analytics within a firm

Phase 3:

Multi-practice templates (employment, PI, wills)

Precedent bank (firm private)

Phase 4:

Dedicated VM per firm option

Client upload portal

Audit logs + stronger security features

DELIVERABLES

Working MVP deployed (central multi-tenant)

Admin login + ability to add firms/users

Matter creation + evidence upload + all analysis features

AI query console working with RAG/embeddings

Export outputs

Full source code + deployment instructions

ACCEPTANCE CRITERIA

The project is accepted when:

A firm account can be created

Matters can be created and searched

Evidence uploads work (PDF, image OCR, WhatsApp txt, email paste)

Chronology builds automatically and is editable

Factor mapping + gap report + comms scanner produce structured outputs

AI query console answers using the matter’s stored data (RAG)

Exports work

TIMELINE EXPECTATION

This is a lean MVP. We expect a fast build like the example provided, which was built in 3 days as we soike to the owner/builder of it. The difference is we are building it for the business side and not the consumer, but the logic is similar.

Please propose a delivery plan with milestones.
We will award Phase 2/3/4 to the same developer if Phase 1 is strong.