CRM

nodyn includes a built-in CRM combining the Knowledge Graph (semantic understanding of people and companies) with DataStore tables (structured deal pipeline and interaction tracking).

How It Works

Knowledge Graph (automatic)
    → Entities: "Lisa Weber works at Acme AG"
    → Relationships: "Acme AG interested in Pro Package"
    → Primary source for "what do I know about X?"

DataStore (agent-driven)
    → contacts: tracked when business-relevant
    → deals: pipeline with stages and values
    → interactions: significant touchpoints logged

Agent decides
    → Recognizes business-relevant person → adds to contacts
    → User discusses opportunity → creates deal
    → Important call/meeting → logs interaction
    → NOT every Telegram message is tracked

The CRM schema is created on first Engine startup. Contact creation is agent-driven, not automatic.

Schema

Three DataStore tables, auto-created:

contacts

Column	Type	Description
name	string	Contact name (unique key, used for upsert)
email	string	Email address
phone	string	Phone number
company	string	Company/organization
type	string	prospect, lead, customer, partner, other
source	string	telegram, email, web, manual
channel_id	string	External ID (e.g. telegram:12345)
language	string	Preferred language (auto-detected from Telegram)
notes	string	Free-text notes

deals

Column	Type	Description
title	string	Deal name (unique with contact_name)
contact_name	string	Associated contact
value	number	Deal value
currency	string	Currency (default: CHF)
stage	string	Pipeline stage
next_action	string	Next step to take
due_date	date	Deadline for next action

Pipeline stages: lead → qualified → proposal → negotiation → won / lost

interactions

Column	Type	Description
contact_name	string	Associated contact
type	string	message, email, call, meeting, note
channel	string	telegram, email, web, manual
summary	string	Interaction summary (auto-truncated)
date	date	Timestamp

Agent-Driven Contact Management

Contacts are NOT auto-created from every message. The agent decides when someone is business-relevant:

• User mentions “Lisa from Acme called about pricing” → agent recognizes business relevance, creates contact + logs interaction
• User sends a casual test message → nothing tracked
• User says “Add Roland as a lead” → agent creates contact explicitly

The Knowledge Graph automatically captures people and companies mentioned in any conversation. The DataStore contacts table is for explicit business tracking (leads, pipeline).

Conversational CRM

You interact with the CRM through natural conversation:

"Zeig mir alle meine Leads"
→ data_store_query on contacts WHERE type = 'lead'

"Erstelle einen Deal für Lisa: Pro Paket, CHF 4800"
→ data_store_insert into deals

"Wie steht meine Pipeline?"
→ data_store_query on deals, aggregation by stage

"Erinnere mich in 3 Tagen bei Roland nachzufragen"
→ task_create with due_date + contact context

"Was weiß ich über Acme AG?"
→ Knowledge Graph: relationships + DataStore: contact + deals + interactions

Knowledge Graph Integration

The CRM data is automatically indexed in the Knowledge Graph:

• Contact entities: “Lisa Weber works at Acme AG”
• Relationships: “Acme AG is a customer since March 2026”
• Context: “Roland prefers email communication”
• Deal history: “v-skin.ch Pro deal at CHF 4800 in negotiation”

This means when you ask “What do I know about Roland?”, nodyn combines structured CRM data with semantic knowledge.

CLI Access

CRM data is stored in standard DataStore tables. Use existing commands:

/data list              # See contacts, deals, interactions tables

Or ask the agent directly — it knows about the CRM tables and can query them.

SDK Usage

import { Engine, CRM } from '@nodyn-ai/core';

const engine = new Engine({});
await engine.init();

const crm = engine.getCRM();

// Agent-driven contact creation
crm.upsertContact({
  name: 'Lisa Weber',
  email: 'lisa@acme.ch',
  company: 'Acme AG',
  type: 'lead',
  source: 'telegram',
});

// Create a deal
crm.upsertDeal({
  title: 'Pro Package',
  contact_name: 'Lisa Weber',
  value: 4800,
  stage: 'proposal',
  due_date: '2026-04-01T00:00:00Z',
});

// Query pipeline
const pipeline = crm.getPipelineSummary();
// → [{ stage: 'proposal', count: 1, total_value: 4800 }]

// Log interaction
crm.logInteraction({
  contact_name: 'Lisa Weber',
  type: 'email',
  channel: 'email',
  summary: 'Sent proposal for Pro Package',
});

Configuration

No configuration needed. The CRM is initialized automatically when the DataStore is available. To disable, don’t initialize the DataStore (set via Engine config).