Every Thunk.AI user can create and save connections to other business applications. They can be managed from the user’s Account page. Once created, these connections can be used within AI tools used in any thunk owned by the user.
Most of the connection types automatically create tool libraries in the user's account with a pre-built set of tools based on the connection. For example, if the user has a Google Drive connection, they automatically have tool libraries to access Google Docs, Google Sheets, and Google Drive.
Sometimes, the user may want to create custom API tools or code-based tools to enhance the capabilities of the AI agents in a thunk. Instead of hard-coding API credentials inside each tool, the user can reference credentials from the connection.
Adding a new Connection
Many application categories are supported and more are being added:
Under each category, some popular applications are listed and you can choose one of them or specify a particular application that is used in your business workflow.
Once the application has been chosen, you get the option to provide a connection Name and a Description. These are important to provide as they inform the AI agents about the information present in the application. For example, if connecting to the Close CRM system, one might chose to say Name: "Close" and Description: "The company CRM including customer leads, accounts, and opportunities".
The connection pane then asks how you plan to connect to the business application.
Most applications will connect using the MCP protocol. It is also possible to connect to business data using a REST API or directly to an underlying SQL database. However, if using these legacy protocols, AI tools are not automatically created. Instead, you will need to manually create tools using the connections.
MCP Connection
The Model-Context-Protocol (MCP) is a standard protocol that all AI agentic platforms use to connect AI agents with external applications. For example, if you have a Close CRM instance and you want an AI agent to be able to read and write data from it:
You would use a CloseCRM-specific MCP server, which exposes a standard API over the CRM data and capabilities that you wish to use. The MCP server has a url endpoint that can be used to access it
You register the MCP server endpoint with your AI agent platform. In Thunk.AI, you do this by adding an MCP connection to your account. The AI agent platform can then act as an MCP client, retrieving the list of available tools and making them available to AI agents to use.
Creating an MCP server from scratch involves writing code and deploying it. Luckily, you do not have to do this. thousands of pre-existing MCP servers have already been created to integrate with most popular systems, and many application vendors are also starting to natively support the MCP protocol. This makes it easy to integrate with a broad range of applications.
There are two kinds of MCP Servers you can use with Thunk.AI:
Native MCP Servers: these MCP servers are natively hosted by the business application. For example, the CloseCRM SaaS application automatically provides an MCP server to allow AI agents to use its data. Many SaaS applications are proceeding down the path of providing native MCP servers. We expect that by mid-2026, every meaningful SaaS application will provide this capability. As this happens, we will automatically include this information into the Thunk.AI platform to minimize what you have to discover. If you want to know if a specific application "<MySaaSApplication>" provides this support, use a Google search for "remote hosted MCP server <MySaaSApplication>".
It is important that this be a Remote MCP service (i.e. accessible via HTTPS)
It is important that the protocol used by "Streamable HTTPS", not the earlier legacy SSE protocol
It is important that this be an MCP service hosted by the SaaS application, not a library package that you are expected to host yourself or install on a local machine.
Via Pipedream: Pipedream is a third-party service unaffiliated with Thunk.AI. They specialize in providing MCP service access to thousands of applications. By integrating with Pipedream, you can create connections to and access data in thousands of applications with minimal effort.
Every MCP server also needs to ensure that access is authenticated. There are different mechanisms used for authentication, just as with any other HTTPS-based API. In most cases, there are only two mechanisms that are relevant:
Using an API key: for many agentic interactions, it is appropriate to get an "API key" of some form and provide it in the HTTPS headers. This is usually appropriate for applications where some kind of "service account" should see all of the data.
Using OAuth: OAuth is a more dynamic way to maintain authentication keys, usually appropriate for applications with end-user accounts and each user sees a different subset of the data.
The official MCP protocol now requires using OAuth, but many of the providers of MCP services provide an API key option. With Thunk.AI, the Pipedream MCP option always uses OAuth which is handled and maintained by the Pipedream service. The Native MCP option supports a variety of options.
Special File System Connections
When you create a new account with Google authentication, the system (and therefore your AI agents) automatically receives a very limited access to your cloud file system. This limited access allows the system to read and write _only_ the files that you explicitly add via the Google Drive file picker UI. This allows for basic testing, but usually is insufficient for fully automated operation.
Likewise, when you create a new account with Microsoft authentication, the system (and therefore your AI agents) receive very limited access to your Microsoft Sharepoint file system.
The cloud file system connection is very important to be able to create and use practical thunks. When building meaningful thunks, it is very common to read files (content folders), read and write spreadsheets (lookup and write relevant information), and create new files (write out summaries or reports). Therefore, we strongly recommend that every user include a Google Drive or Microsoft Sharepoint file connection with read and write permissions. To add these permissions, go to your Account -> Connections page, choose the specific connection, and click on the appropriate button to add the necessary permissions.
Authenticated Website Connection
As an integral part of the standard tool libraries in Thunk.AI, there is an open_url tool used to read web pages. When an AI agent uses this tool, it uses a hosted web browser to load the page and read its content. While this works with any public web page, it does not work by default for web pages that require authentication.
We generally discourage customers from trying to access information via web pages when there is an API alternative. Many web pages are hosted on infrastructure that tries to block or limit automated access (eg: captchas, rate limits, IP blocks, etc). When it comes to authenticated web site connections, there are further issues related to the expiry of credentials. For these reasons, although the Thunk.AI platform has implemented support for access to such authenticated web sites, we have disabled this feature. Please contact [email protected] if you believe this is absolutely essential for your scenario.