With Extensions, you can connect your AI Agent to external services. This allows the agent to directly access data or trigger actions in other tools — without you having to copy or switch manually.
What is an Extension?
An Extension connects your AI Agent to an external system via an interface (API or MCP). The agent can then read data (e.g. retrieve applicants from an ATS) or execute actions (e.g. update a status).
You can add up to 10 Extensions per account.
Available Types
MCP — Model Context Protocol: Connects the agent to an MCP server (e.g. Notion, Memberspot Academy)
API — Direct HTTP connection to a REST API (e.g. Recruitee, custom systems)
Notion — Preconfigured Notion integration via OAuth
Intercom — (coming soon)
Creating an Extension
Go to AI Agents → "Extensions" tab
Click "+ Add Extension"
Choose the extension type (MCP, API, Notion ...) → Continue
Fill in the form:
Name — e.g. "Notion" or "Recruitee"
Description — What does this Extension do?
URL — Endpoint of the MCP server or API
Auth Type — None / API Key / Bearer Token (depending on the service)
Click "Create Extension"
Note for OAuth: For services like Notion, after creating the Extension you will be prompted to click "Connect" and authorize via OAuth. Only then will the agent recognize the available actions.
Managing Actions
Once the Extension is connected, Memberspot automatically displays all recognized actions — split into:
Read Tools — The agent can retrieve data (GET requests)
Write Tools — The agent can modify data (PATCH/POST requests)
For each action you can individually control whether the agent is allowed to use it and whether it should ask the user for confirmation first.
Extension or Knowledge Tool — What to Use When?
Both options extend your AI Agent, but they work in fundamentally different ways:
Extension | Knowledge Tool | |
How | Live query in real time | Periodic sync into vector database |
Data | Always up to date | Snapshot at the time of last sync |
Can write? | Yes | No |
Best for | Actions, live data | Static knowledge, documents |
When to Use an Extension?
Use an Extension when:
The agent should actively do something (change a status, update a record, send an email)
You need always up-to-date data (e.g. Show me today's applicant pipeline)
The data changes frequently and an outdated snapshot would be a problem
You want to retrieve a specific single record (e.g. an applicant by their ID)
Example: Recruitee Extension — the agent retrieves applicants live and moves them to the next stage.
When to Use a Knowledge Tool?
Use a Knowledge Tool when:
It is about documents, FAQs, guidelines or knowledge base content
The content rarely changes and a daily/weekly sync is sufficient
The agent should search semantically (e.g. What does our handbook say about vacation policies?)
You want to make large amounts of text or many documents accessible
Example: HR documents as a knowledge source — the agent searches contracts and policies via vector search.
Important: Performance with Large Datasets
Extensions make live API requests — the more data exists in the external system, the longer the search takes. If an agent has to iterate through thousands of records in an API to find the right answer, it becomes slow and error-prone.
Knowledge Tools solve exactly this: data is pre-indexed and vectorized — the search is then lightning fast because the agent no longer has to scan the API, but searches directly in the index.
Rule of Thumb
Few, targeted records + executing actions → Extension
Large amounts of data / many documents to search → Knowledge Tool
Combining both is possible
Combine both: In practice, both complement each other. An onboarding agent could know the handbook via a Knowledge Tool and create new employees in the HR system via an Extension.