The short answer is:
API = lets software communicate with Loft47 programmatically.
MCP = gives AI assistants a standardized way to use Loft47 tools.
They can work together, but they serve different purposes.
An API (Application Programming Interface) allows another application or system to communicate with Loft47 programmatically.
For example, a company could build its own application or AI Agent that uses the Loft47 API to:
Retrieve deals
Retrieve agents and contacts
Get commission and split information
Retrieve transaction or payout information
Create or update records, if supported
Run reports
Automate workflows
For example:
“Every morning, pull all pending deals from Loft47 and compare them against our CRM.”
Their application could use the Loft47 API to retrieve the data automatically.
The flow would be:
Their Application / AI Agent → Loft47 API → Loft47
An API does not provide the AI itself. It simply provides programmatic access to Loft47.
MCP (Model Context Protocol) is a standardized way for AI applications to connect to and use external tools and data.
With Loft47 MCP, a user can connect Loft47 to an AI assistant such as Claude and use natural-language prompts.
For example:
“Show me all deals closing this week.”
Claude can understand the request, determine which Loft47 tool it needs, retrieve the information through MCP, and respond to the user.
The flow is:
User → Claude → Loft47 MCP → Loft47
This is especially useful for users who want to interact with Loft47 through AI without needing to write code.
| API | MCP + AI | |
|---|---|---|
| Main purpose | Software-to-software communication | AI-to-software communication |
| Typical user | Developer / application | End user |
| AI included? | No | Yes, through the AI platform |
| Natural-language interaction | Not by itself | Yes |
| Can build custom applications? | Yes | Less direct |
| Requires Claude? | No | Typically yes for a Claude-based setup |
| Can work with different AI platforms? | Yes | Depends on the AI platform's MCP support |
Because they may not be using Claude.
A company might be building its own AI Agent and want it to connect to several different systems:
Their AI Agent
Loft47 API → deal and commission data
QBO API → accounting information
CRM API → customer information
Email API → send emails
Calendar API → schedule meetings
AI/LLM API → reasoning and conversation
The AI Agent can then coordinate all of these systems.
For example, a user could ask:
“Find all deals closing this week, check whether they've been posted to QBO, and email me a list of anything missing.”
The AI Agent could:
Use the Loft47 API to find the deals.
Use the QBO API to check posting status.
Analyze the results.
Use an Email API to send the report.
The user doesn't necessarily need Claude at all.
Yes.
Think of it this way:
API = the plumbing
MCP = the standardized AI interface
Claude = the AI that decides what to do
For example, Loft47 could have:
Loft47 system → Loft47 API → Loft47 MCP → Claude
The MCP layer can expose Loft47 capabilities as AI-friendly tools, while the API provides the underlying programmatic connection.
So MCP doesn't necessarily replace the API.
An API is AI-platform agnostic.
If a company has access to the Loft47 API, they could potentially build an AI Agent using:
ChatGPT / OpenAI
Claude / Anthropic
Google Gemini
Microsoft Copilot
Glean
Guru
A custom AI application
Their own AI Agent
The architecture could look like:
Their AI Agent → Loft47 API → Loft47
Whereas your current Loft47 AI experience is:
Claude → Loft47 MCP → Loft47
This is why APIs can be valuable for Loft47: they allow Loft47 to become a data and workflow source for many different applications and AI platforms, rather than limiting access to one particular AI assistant.
You want to use AI with Loft47 and interact with it through natural language.
For example:
“Show me my pending deals.”
“Which agents are over their cap?”
“Find deals with missing payouts.”
You are building your own application, automation, or AI Agent and want programmatic access to Loft47.
For example:
“Build an AI Agent that connects Loft47, QBO, our CRM and email system.”
You're building a sophisticated AI system and want to expose Loft47's capabilities to an AI through MCP while also using APIs for other applications and services.
API: “I want my software to talk to Loft47.”
MCP: “I want my AI to use Loft47.”
Claude + MCP: “I want to talk to Loft47 using natural language through Claude.”
API + AI Agent: “I want to build my own AI system that can use Loft47 along with other systems.”