Product Introduction
- Definition: Relaticle is an open-source, self-hosted Customer Relationship Management (CRM) platform built on a modern Laravel/PHP stack with a PostgreSQL database. It is engineered from the ground up to be "agent-first," meaning its core architecture is designed for seamless integration with both built-in and external AI agents.
- Core Value Proposition: Relaticle exists to unify human and AI-powered workflows within a single, data-sovereign CRM. It solves the problem of "bolted-on" AI in traditional CRMs by providing deep, native integration through a built-in AI chat assistant and a comprehensive MCP (Model Context Protocol) server, all while giving users complete control over their data through open-source self-hosting.
Main Features
- Built-in AI Chat Assistant ("Rela"): This feature provides a conversational interface directly within the CRM UI. Users can ask natural language questions about their data (e.g., "What's overdue this week?"), create records, or request summaries. Technically, it uses a configurable AI provider endpoint (Claude, Gemini, OpenAI-compatible) to process queries. A critical security mechanism is its "proposal" system: any write or destructive action generates a preview card requiring explicit user approval before execution, with a 5-second undo function.
- MCP Server for External AI Agents: Relaticle includes a first-party MCP server exposing 37 tools over OAuth 2.0. This allows any MCP-compatible client (e.g., Claude Desktop, Cursor, ChatGPT) to perform authenticated CRUD operations, pipeline analysis, and activity history queries directly on the CRM data. The agent's tool schema automatically includes the workspace's custom fields, ensuring external AI has full context of the customized data model.
- Customizable Data Model with 22 Field Types: Beyond standard CRM fields, Relaticle supports advanced types including entity relationship fields (linking records), conditional visibility logic, and per-field encryption. This allows teams to model their unique business processes without requiring database migrations or code changes, all managed through the administrative UI.
- Self-Hosted Deployment & AGPL Licensing: The entire codebase is licensed under the AGPL-3.0, allowing for full audit, modification, and private deployment. It is distributed as a Docker container (with over 28,000 pulls) and can be deployed on any PHP 8.5+ environment. This feature ensures data never leaves a user's own infrastructure, and local inference can be used with solutions like Ollama.
Problems Solved
- Pain Point: Disconnect between AI agents and core business systems. Many teams use powerful AI agents that lack secure, structured access to live CRM data, leading to manual copy-pasting or insecure API workarounds.
- Target Audience: Technical founders, product-led sales teams, DevOps engineers, and AI/ML practitioners who require a CRM that integrates natively with their AI toolchain. It's ideal for privacy-conscious organizations (e.g., legal, healthcare, finance) and those wanting to avoid per-seat SaaS pricing models.
- Use Cases: A salesperson using Claude Desktop can directly create a contact and link it to a company in Relaticle via natural conversation. A manager can ask the in-app chat to "list all deals in the proposal stage over $50k" and approve a batch status update. A development team can self-host Relaticle on their private cloud, using local LLMs for full data privacy.
Unique Advantages
- Differentiation: Unlike SaaS CRMs like HubSpot or Salesforce that add AI features as proprietary, vendor-locked extensions, Relaticle is open-source and provides deep, standardized integration (via MCP) with any AI agent. Its pricing is a flat fee per workspace, not per user, challenging the industry-standard per-seat model.
- Key Innovation: The "agent-first" architecture is the core innovation. By baking the MCP server and a principled, approval-based AI assistant directly into the product's foundation, Relaticle treats AI agents as first-class users with secure, scoped access, rather than an afterthought. The automatic inclusion of custom fields in the MCP schema is a significant technical detail that eliminates configuration overhead.
Frequently Asked Questions (FAQ)
- Is Relaticle secure for sensitive CRM data? Yes. The self-hosted option ensures all data remains on your infrastructure. The built-in AI chat uses your configured API keys, and its approval gate prevents unauthorized changes. The platform features 5-layer authorization (role-based permissions, workspace isolation, etc.) and supports per-field encryption for sensitive data.
- How does Relaticle's MCP integration work with ChatGPT or Claude? You connect your MCP client (e.g., Claude Desktop) to Relaticle's MCP server using OAuth 2.0. Once connected, the AI agent has access to 37 tools like
list_companiesorcreate_contact. When you ask the agent a question about your sales pipeline, it can use these tools to fetch live data from your Relaticle instance and operate on it with your permission. - What is the difference between the built-in AI chat and external MCP agents? The built-in chat ("Rela") is a controlled interface within the Relaticle web app with strict approval workflows, ideal for direct user interaction. External MCP agents provide flexibility to use your preferred AI interface (e.g., IDE, dedicated agent app) with the same powerful data access, enabling more automated and customized workflows.
- Can I migrate my data from Salesforce or HubSpot to Relaticle? Yes. Relaticle includes CSV import functionality with column mapping and validation, which is the standard method for migrating from other CRM platforms. As an open-source solution, you also have the option to build direct data connectors if needed.
- What does "free forever self-hosted" mean for the Relaticle CRM? It means the core AGPL-3.0 licensed software has no cost or user limit for self-hosting. You can deploy it on your own servers indefinitely. The company generates revenue through its managed cloud hosting option (
app.relaticle.com), which uses simple, flat per-workspace pricing.
