customer-support-agent

Customer support agent

First-line AI customer support on DNotifier: strong system + knowledge prompts, RAG via the knowledge base, and session-aware chat.

What you get

• System prompt — tone, steps, safety, escalation rules for proper support
• Knowledge prompt — grounds answers in indexed support docs when RAG is on
• addKnowledgeBase — indexes a document through DNotifier addDocument
• Sample FAQs — password reset, refunds, shipping, contact hours
• CLI chat — ask questions, seed/add docs, toggle KB, fetch history

Run examples

Create an app at app.dnotifier.com, then copy .env.example → .env and set DNOTIFIER_APP_ID / DNOTIFIER_SECRET.

TypeScript (Node)

cd typescript
cp .env.example .env
npm install
npm start

Python

cd python
cp .env.example .env
python3 -m venv .venv
source .venv/bin/activate   # Windows: .venv\Scripts\activate
pip install -r requirements.txt
python main.py

Dart

cd dart
cp .env.example .env
dart pub get
dart run bin/main.dart

Once running, try:

you> How do I reset my password?
agent> …

you> What is your return window?
agent> …

CLI commands: /seed, /add Title | content, /kb on|off, /history, /help, /quit.

DNotifier building blocks

Capability	API
Connect	connect() with transport: "http"
Answer	sendAI / send_ai with messages, useKnowledgeBase, sessionId, saveHistory
Index docs	addDocument / add_document (wrapped as addKnowledgeBase)
History	fetchAIHistory / fetch_ai_history
Observability	logs: true → dashboard sessions

Use HTTP for AI + RAG request/response. Use WebSocket when you need live multi-user messaging.

Implement it yourself

1. Create an app at app.dnotifier.com (AI-enabled plan).
2. Install the SDK for your language.
3. Connect with appId, secret, transport: "http".
4. Call addKnowledgeBase / addDocument with your FAQs and policies (recordId required in Python and Dart).
5. Call sendAI with the support system prompt (+ knowledge prompt when RAG is on), useKnowledgeBase: true, and a stable sessionId for follow-ups.
6. Optionally enable logs: true and inspect the run in the dashboard.

Languages

Language	Status	Path	Run
TypeScript (Node)	Ready	typescript/	npm start
Python	Ready	python/	python main.py
Dart	Ready	dart/	dart run bin/main.dart

Docs

• Build an AI assistant
• Knowledge base / RAG
• AI sessions