Skip to content

Quick Start

Prerequisites#

  • Node.js 18+ — Required for the CLI and SDK
  • npm, pnpm, or yarn — Any package manager works
  • An Cognest API keyGet one here

Install the CLI#

npm install -g @cognest/cli

Verify the installation:

terminal
cognest --version
# Cognest CLI v2.4.1

Initialize Your Project#

terminal
cognest init my-assistant
cd my-assistant

This creates a project with the following structure:

project structure
my-assistant/
├── cognest.config.yaml    # Main configuration
├── .env                    # Environment variables (credentials)
├── skills/                 # Custom skill definitions
│   └── hello-world.ts      # Example skill
├── package.json
└── tsconfig.json

Configure Your Assistant#

Open cognest.config.yaml and set your API key and preferred model:

cognest.config.yaml
# cognest.config.yaml
assistant:
  name: "My Assistant"
  model:
    provider: openai
    model: gpt-4o
    temperature: 0.7
  system_prompt: |
    You are a helpful personal assistant. Be concise,
    friendly, and proactive about following up on tasks.

integrations:
  whatsapp:
    enabled: true
    credentials:
      phone_number_id: ${WHATSAPP_PHONE_ID}
      access_token: ${WHATSAPP_ACCESS_TOKEN}

skills:
  - email-triage
  - calendar-sync

Add Your First Integration#

terminal
# Add WhatsApp integration
cognest integration add whatsapp

# The CLI will prompt you for credentials:
# ✓ WhatsApp Phone Number ID: ••••••••
# ✓ WhatsApp Access Token: ••••••••
# ✓ Webhook Verify Token: ••••••••
#
# ✓ WhatsApp integration configured successfully

Multiple integrations

You can add as many integrations as you need. Run `cognest integration add <name>` for each platform. See the Integrations section for the full list.

Start the Development Server#

terminal
cognest dev

# ✓ Loading configuration...
# ✓ Connecting integrations...
#   ✓ WhatsApp connected (webhook: http://localhost:3100/webhook/whatsapp)
# ✓ Loading skills (2 active)...
# ✓ Think Engine ready (gpt-4o)
#
# 🦀 Assistant "My Assistant" running at http://localhost:3100
#    Dashboard: http://localhost:3100/dashboard

Your assistant is now running locally. Send a message on WhatsApp to your configured phone number and watch the Think Engine process it in real time.

Deploy to Production#

When you're ready to go live, deploy to Cognest with a single command:

terminal
cognest deploy

# ✓ Building project...
# ✓ Uploading to Cognest...
# ✓ Provisioning infrastructure...
# ✓ Connecting integrations...
#
# 🚀 Deployed successfully!
#    URL: https://my-assistant.cognest.run
#    Dashboard: https://app.cognest.ai/my-assistant

Self-hosting? See the Deployment Options guide for Docker, Kubernetes, and bare-metal deployment instructions.

Next Steps#