Skip to main content
This walks you from a clone to a grounded answer. For the full prerequisite matrix and production hardening, see Installation.
AskMyDocs is a Laravel 13 application. You need PHP 8.3+, Composer, Node 20+, and a PostgreSQL 15+ database with the pgvector extension. The Quickstart uses OpenAI for chat + embeddings; any supported provider works (see AI providers).
1

Clone and install

git clone https://github.com/lopadova/AskMyDocs.git
cd AskMyDocs
composer install
npm install && npm run build
2

Configure the environment

Copy the example env and set your database + one AI provider key.
cp .env.example .env
php artisan key:generate
Then edit .env:
DB_CONNECTION=pgsql
DB_HOST=127.0.0.1
DB_PORT=5432
DB_DATABASE=askmydocs
DB_USERNAME=postgres
DB_PASSWORD=secret

AI_PROVIDER=openai
AI_EMBEDDINGS_PROVIDER=openai
OPENAI_API_KEY=sk-...
Embedding dimensions are part of the contract. The stock schema is sized for a 1536-dim model (openai/text-embedding-3-small). Switching to a different-dimension model means migrating the vector(N) columns, flushing the cache, and re-indexing — see Installation.
3

Migrate the database

pgvector must be installed in PostgreSQL first (CREATE EXTENSION vector;). Then run the migrations:
php artisan migrate
4

Ingest some documents

Point the folder ingester at a directory of markdown and assign a project key. Ingestion is idempotent — re-running on identical bytes is a no-op.
php artisan kb:ingest-folder docs/ --project=handbook
The same execution path is available over HTTP (POST /api/kb/ingest, Sanctum-protected, batch ≤ 100) — see Architecture overview.
5

Ask a grounded question

Use the stateless JSON API:
curl -X POST https://your-host/api/kb/chat \
  -H "Authorization: Bearer <token>" \
  -H "Content-Type: application/json" \
  -d '{"question": "What is our remote-work policy?", "project_key": "handbook"}'
The response carries the answer, the citations that grounded it, and a meta block (provider, model, latency, retrieved-chunk count). Or open the React chat UI at /app/chat for the streaming experience with inline citations and suggested follow-ups.

​
What just happened

  1. Your markdown was chunked (section-aware), embedded, and upserted into PostgreSQL on the idempotency tuple (project_key, source_path, version_hash).
  2. Your question was embedded and run through hybrid retrieval — vector + full-text — then fused by the reranker.
  3. The grounded context was composed into a prompt and sent to your provider via AiManager, and the answer came back with citations to the exact chunks.

​
Next steps

Core concepts

Understand the canonical layer and the knowledge graph.

Connectors

Pull from Google Drive, Notion, Confluence, Jira and more.

Admin panel

Dashboards, RBAC, the KB explorer, logs, and maintenance.