trycompai/crm: The Open-Source, Agentic-First CRM Revolutionizing Sales Workflows

Customer Relationship Management (CRM) systems have long been the backbone of sales operations. But traditional CRMs are reactive — they wait for humans to enter data, update fields, and trigger workflows. In a world where AI agents are becoming digital coworkers, an agentic-first CRM flips the script. Enter trycompai/crm: an open-source, agentic-first CRM designed from the ground up for autonomous AI collaboration.

In this article, we'll explore what makes trycompai/crm different, its core architecture, how to get started, and why this shift matters for modern sales teams.

What Is an Agentic-First CRM?

An agentic-first CRM isn't just a CRM with a few AI features bolted on. It's a system where AI agents are first-class citizens. Instead of a human manually logging every call, email, or meeting, AI agents actively monitor, update, and act on data across the entire customer lifecycle.

Agents in trycompai/crm can:

• Automatically enrich lead records from public sources
• Draft personalized follow-up emails and schedule meetings
• Route leads to the right sales rep based on intent and fit
• Predict churn risk and alert account managers
• Execute complex workflows without human supervision

The fundamental difference is that agents aren't just suggesting actions — they're performing them. This transforms the CRM from a database of record into an autonomous operations hub.

Why Open Source Matters for Agentic CRMs

Proprietary CRMs like Salesforce and HubSpot are adding agentic features, but they're locked ecosystems. With an open-source, agentic-first CRM, you retain full control over:

• Data privacy: Run it on your own infrastructure; your customer data stays yours.
• Agent behavior: Modify the prompt templates, tools, and decision logic.
• Integrations: Connect to your internal APIs and data lakes without waiting for a vendor.
• Cost: Avoid per-seat and per-agent licensing fees.

For companies building custom AI pipelines, open source is the only reasonable foundation.

Core Features of trycompai/crm

trycompai/crm is built with a microservices architecture, using modern TypeScript, Node.js, and a React frontend. Here are its standout capabilities:

1. Agent-Aware Data Model

Every CRM entity (Contact, Account, Lead, Opportunity) has an associated agent_metadata object. This stores not just historical human edits, but also agent reasoning traces — what the agent observed, decided, and acted upon. This transparency is crucial for auditability.

2. Workflow Automation Engine

The built-in workflow engine lets you define triggers (e.g., "lead score changes") and agent actions (e.g., "send a nurturing email"). You can chain multiple actions with conditions, loops, and human approval steps.

3. Unified Agent API

trycompai/crm exposes a REST and GraphQL API that agents can call to read and write data. More importantly, it provides an event bus — agents can subscribe to domain events and react in real time. For example, an agent could listen for "new deal created" and immediately start assembling a proposal.

4. Human-in-the-Loop Approval Gates

Full autonomy isn't always desirable. trycompai/crm lets you set approval thresholds. If an offer discount exceeds 20%, the agent must wait for a sales manager to approve before proceeding. This balances efficiency with risk management.

5. Observability Dashboard

The dashboard provides an agent activity feed, showing every action taken by every agent. You can filter by agent, entity, or action type, and even replay a step-by-step trace for debugging.

Getting Started with trycompai/crm

Getting up and running is straightforward. The project ships with a docker-compose.yml that spins up the entire stack: Postgres, Redis, the API, the web app, and a sample agent worker.

# Clone the repository
git clone https://github.com/trycompai/crm.git
cd crm

# Copy environment configuration
cp .env.example .env

# Start all services
docker-compose up -d

Once the containers are running, navigate to http://localhost:3000 to access the dashboard. The default login credentials are provided in the .env.example file.

Configuring Your First Agent

Agents are defined as YAML files. Here's a minimal example that creates a lead enrichment agent:

name: lead-enricher
model: gpt-4o
schedule: "*/5 * * * *"   # run every 5 minutes

steps:
  - action: query
    # Fetch all leads with missing company size
    api: internal
    endpoint: /leads
    params:
      filter: "company_size IS NULL"
  - action: api_call
    # Use an external enrichment service
    url: "https://api.enrich.example.com"
    headers:
      Authorization: "Bearer ${ENRICH_API_KEY}"
    body:
      domain: "{lead.website}"
  - action: update
    entity: lead
    fields:
      company_size: "${result.employees}"
      industry: "${result.industry}"

Add your agent YAML files to the agents/ directory, and the agent worker will automatically load them.

Building Custom Agent Tools

trycompai/crm lets you extend agents with custom tools using a simple function interface. If you have an internal pricing tool, expose it as a tool for agents:

// tools/pricing.ts
import { Tool } from '@trycompai/agent-sdk';

export const pricingTool: Tool = {
  name: 'get-pricing',
  description: 'Fetch tiered pricing for a product or service',
  parameters: {
    type: 'object',
    properties: {
      productId: { type: 'string' }
    }
  },
  async execute({ productId }) {
    // Call your internal pricing microservice
    const pricing = await fetch(`/api/pricing/${productId}`).then(r => r.json());
    return pricing;
  }
};

Then register it in your agent configuration:

name: sales-assistant
tools:
  - get-pricing
  - send-email
  - create-task

Real-World Use Cases

Lead Qualification and Routing

An agent can analyze incoming web forms, score leads using your ICP criteria, and assign them to the appropriate representative — all within seconds of the lead submission.

Account Health Monitoring

Agents can scan support tickets, usage logs, and payment history to detect early signs of churn. When a risky account is detected, the agent opens a task for the customer success manager and drafts a proactive outreach message.

Meeting Preparation Briefs

Before every sales meeting, an agent collates the account's recent interactions, open opportunities, and relevant news, then generates a one-page brief delivered to the rep's inbox.

Comparing trycompai/crm with Traditional CRMs

Feature	Traditional CRM	trycompai/crm
Data entry	Manual	Agent-automated
Workflow triggers	Basic rules	AI decision-making
Extensibility	Vendor app store	Open-source + API
Observability	Activity logs	Full agent traces
Cost model	Per-seat license	Self-hosted free

The key differentiator is that in traditional CRMs, automation is rule-based and deterministic. trycompai/crm introduces probabilistic reasoning into every workflow, allowing agents to handle nuanced scenarios like understanding a customer's emotional tone or deciding the best time to send a follow-up.

Security and Data Governance

When you let AI agents operate on customer data, security becomes even more critical. trycompai/crm addresses this in several ways:

• Role-based access control: Agents have their own API keys with restricted scopes.
• Row-level security: Agents can only access records they're permitted to see.
• Full audit trail: Every agent action is logged immutably.
• Privacy modes: You can configure agents to redact PII before sending data to external LLM APIs.

Roadmap and Community

trycompai/crm is under active development. The roadmap includes:

• Multi-agent orchestration with agent-to-agent communication
• Native Slack and WhatsApp integrations
• An embeddable chat widget that lets prospects interact with an agent directly
• Fine-tuned small language models for on-premise inference

The open-source community is growing, and contributions are welcome in areas like connector development, agent templates, and documentation.

Conclusion

trycompai/crm represents a bold step forward: a CRM that doesn't just store your relationships but actively nurtures them. By building an open-source, agentic-first platform, it empowers teams to automate the mundane while keeping humans in control of the strategic decisions.

Whether you're a startup looking to punch above your weight or an enterprise seeking to reduce CRM admin burden, trycompai/crm offers a modern, transparent, and future-proof foundation for your sales operations.

Ready to let your CRM work for you? Clone the repo, spin up your first agent, and join the agentic revolution.