Overview
Softools supports connecting compatible AI assistants using Model Context Protocol (MCP), allowing you to ask questions about your Softools Apps and Records using natural language.
Instead of opening individual Apps and reports to find information, you can ask an AI assistant to retrieve, organise and analyse your Softools data.
For example, you could ask:
- Which Projects have the highest priority and need attention?
- What are the biggest Risks affecting a particular Project?
- Which Actions are overdue?
- Which Projects am I responsible for?
- Can you find the Project Charter and provide a link to it?
- Can you provide direct links to the relevant Softools Records?
The assistant can retrieve information from multiple Apps, helping you understand related Projects, Actions, Risks and other information in a single conversation.
Softools MCP is read-only. AI assistants can retrieve and search information but cannot create, update or delete Softools Records through the MCP connection.
The information available is controlled by published MCP-enabled Views and each user's existing Softools permissions.
Contents
- How MCP Works
- Preparing Softools for MCP
- Choose Your AI Assistant
- What Can You Ask an AI Assistant?
- Security and Permissions
- Troubleshooting
How MCP Works
Model Context Protocol (MCP) is a standard that allows AI assistants to connect to external systems and use the tools and information those systems make available.
Softools provides an MCP server that compatible AI assistants can connect to.
There are three main stages to using Softools MCP:
- Prepare the information. An App Builder creates and publishes MCP-enabled Views containing the Fields that the AI assistant needs.
- Approve the AI assistant. An Advanced App Builder approves the connection to the Softools workspace.
- Connect your account. Users authorise access using their own Softools accounts and can then ask questions about the information available to them.
Once connected, the AI assistant can use Softools MCP tools to retrieve and search Records from the published Views.
Users do not need to be App Builders to ask questions about their Softools data.
Preparing Softools for MCP
Softools uses published Views to control which information is available through MCP.
An App Builder can configure Views to expose useful Fields, such as Project Name, Status, Priority, Delivery Date and Project Lead.
They can also:
- Add Field descriptions to help AI assistants interpret values correctly.
- Enable Data Lens for meaning-based searches.
- Include references to related Apps, allowing assistants to work across Projects, Actions and Risks.
- Include Document Fields to make document information and temporary links available.
- Include direct Record URL Fields so assistants can return links that open Records in Softools.
For example, a Projects View could be combined with separate Actions and Risks Views to help an assistant identify delivery concerns and the work required to address them.
The Softools MCP server URL is available to App Builders under App Studio > Configure > MCP connections.
For step-by-step instructions, see Preparing Softools Apps and Views for MCP.
Choose Your AI Assistant
Softools provides connection options for Claude, Microsoft Copilot Studio and other compatible AI assistants.
An Advanced App Builder can manage these connections under App Studio > Configure > MCP connections.
Choose the guide that matches the AI assistant you want to use.
Claude
Claude has a dedicated connection option in Softools, using its published identity.
An Advanced App Builder approves Claude for the workspace, after which individual users can add Softools as a connector and authorise access with their own Softools accounts.
Follow these guides:
- Approving Claude for Softools MCP — For Advanced App Builders approving Claude.
- Connecting Claude to Softools — For users adding the Softools connector in Claude.
Microsoft Copilot Studio
Microsoft Copilot Studio allows organisations to build AI agents that can retrieve information from Softools through MCP.
An administrator configures the agent's MCP connection and authentication, and an Advanced App Builder approves the connection in Softools.
The organisation can then publish the agent through its chosen supported channels, including Microsoft 365 Copilot or Microsoft Teams.
Follow these guides:
- Connecting Microsoft Copilot Studio to Softools — For configuring and approving the MCP connection.
- Using a Softools AI Assistant in Microsoft Teams — For users accessing an organisation-provided agent through Teams.
ChatGPT and Other AI Assistants
For compatible AI assistants without a dedicated connection option, Softools provides Manual / another AI assistant.
This allows an Advanced App Builder to approve the client using its OAuth connection details.
Follow this guide:
- Connecting Other AI Assistants to Softools Using MCP — Includes a worked example using ChatGPT.
The connection options available in an AI assistant may depend on its subscription, configuration and support for MCP.
What Can You Ask an AI Assistant?
Once connected, you can ask questions about the Softools information available to your account.
Retrieve and Organise Records
For example:
Show all Projects with their RAG Status, Priority and Delivery Date. Group them by Status, with Red first, and sort each group by Delivery Date.
The assistant can retrieve Project Records, apply the requested grouping and present the results in a structured response.
Structured Query tools are particularly useful for filtering, sorting and retrieving complete lists of matching Records, including additional pages where required.
Analyse Information Across Apps
If your organisation has published Projects and Risks Views, you could ask:
For New Product Research & Development Facility, what are the biggest Risks and in which order should they be addressed?
The assistant can retrieve the linked Risk Records, compare their recorded scores and recommend an order of attention.
In this example, the assistant identifies two Risks, ranks them using their recorded Risk Scores and suggests actions to address them.
The example uses an agent in Microsoft Teams, but the underlying Softools MCP tools can also be used by other compatible AI assistants.
Search for Relevant Information Using Data Lens
Where Data Lens is enabled, you can ask questions based on meaning rather than exact Field values.
For example:
Search my Projects for delivery delays, resource shortages or other significant Risks.
The assistant can use Data Lens Search to find Records whose information is relevant to the question, even when the wording differs from the search terms.
Data Lens Search returns ranked matches rather than an exhaustive list. For precise filtering, sorting or complete lists, use structured Query tools.
Retrieve and Analyse Documents
Softools MCP can also make information about documents associated with Records available to AI assistants.
When an App Builder includes a Document Field in an MCP-enabled View, the assistant can retrieve document information, including the filename and a temporary link to the file.
For example:
Give me the Project Charter link for the ERP Upgrade Project.
The assistant can retrieve the document information from Softools and provide a link to the file.
In this example, Claude retrieves a link to the ERP Upgrade Project Charter and identifies when the temporary link expires. The document is then attached to the conversation for further analysis.
Once the document is available to the assistant, you could ask:
Summarise the key objectives and delivery milestones in the Project Charter.
Compare the Project Charter with the current Project Status, Actions and Risks in Softools. Are we on track to meet the original objectives?
This allows users to combine information from project documents with current Softools Records.
Important: An assistant may be able to retrieve the document details and temporary link without being able to open and read the file itself. Some AI platforms restrict access to external document storage domains.
In these cases, users can download the document and, where permitted by their organisation, upload it to the AI conversation for analysis.
Document links may expire and should not be treated as permanent links.
For more information about configuring Document Fields, see Preparing Softools Apps and Views for MCP.
Open Records Directly in Softools
Where an MCP-enabled View includes a direct Record URL Field, you can ask:
Show my Red Projects and include a link to each Record in Softools.
The assistant can return clickable links alongside the retrieved information.
This allows you to move from an AI-generated summary directly to the underlying Softools Record.
For more practical examples, see Using AI Assistants with Softools Data.
Security and Permissions
Softools MCP uses several controls to manage access to information:
- Connection approval — An Advanced App Builder approves which AI assistants can connect to the workspace.
- Individual authorisation — Users sign in with their own Softools accounts and grant read access.
- Published Views — App Builders determine which Fields and information are exposed through MCP.
- Existing permissions — Users can only retrieve information available to their Softools accounts.
- Read-only tools — The MCP connection does not provide tools for creating, updating or deleting Records.
Information retrieved by an external AI assistant, including documents uploaded for analysis, may also be subject to that provider's data-handling and retention policies.
Only connect AI assistants approved by your organisation, and review their responses against the underlying Softools information before making important decisions.
For detailed guidance, see Security Considerations When Connecting AI Assistants to Softools.
Troubleshooting
I cannot find the MCP server URL
The URL is available to App Builders in App Studio > Configure > MCP connections.
Ask your App Builder or administrator to provide it if you do not have access.
My AI assistant cannot connect to Softools
Check that the assistant has been approved for the workspace and that you have completed the Softools sign-in and authorisation process.
Follow the relevant connection guide for detailed instructions.
An App or View is missing
Check that the required View has been enabled for MCP and that the App has been published.
Also confirm that your Softools account has permission to access the information.
Newly published information is not available
The AI assistant may need to refresh its MCP tools after an App Builder adds or changes a published View.
Refresh or reconnect the Softools connection and start a new conversation if necessary.
The assistant returns incomplete or unexpected information
Check that the relevant Fields are included in the published Views and have clear descriptions.
Use structured Query tools for complete lists of matching Records, rather than relying on ranked Data Lens Search results.
Ask the assistant to identify the Records and Fields used to support its response.
The assistant can find a document but cannot read it
The MCP response may include the document's filename and temporary link, but the AI platform may be unable to open the file.
Open the document link yourself and, where permitted, upload the document to the AI conversation for analysis.
If the link has expired, request a fresh link through the assistant.
Related Articles
Preparing and Connecting
- Preparing Softools Apps and Views for MCP — Configure the information available to AI assistants.
- Approving Claude for Softools MCP — Approve Claude for your workspace.
- Connecting Claude to Softools — Connect your Softools account to Claude.
- Connecting Microsoft Copilot Studio to Softools — Configure a Copilot Studio agent.
- Connecting Other AI Assistants to Softools Using MCP — Connect ChatGPT or another compatible assistant.
Using MCP and Managing Access
- Using a Softools AI Assistant in Microsoft Teams — Use an organisation-provided agent through Teams.
- Using AI Assistants with Softools Data — Practical examples of queries, searches, document links and Record links.
- Security Considerations When Connecting AI Assistants to Softools — Understand permissions, authentication and data handling.
For a practical demonstration, see the Softools Community post
Comments
0 comments
Article is closed for comments.