The open beta for Procore Copilot is available to customers in the United States using Procore in English who license the required toolset. Follow the steps in Enable Copilot to start participating.
If you are using Procore in the United States in English, but do not see the option to opt in to the beta, it is likely because your company does not license the tool package required to be eligible to participate in the beta at this time.
Customers using Procore in English in Canada, the United Kingdom and Ireland, Australia, and New Zealand are eligible to participate in a Closed Beta. Contact your Procore point of contact to request access to the closed beta.
Table of Contents
Copilot is Procore's AI assistant tool. Copilot can respond to questions asked in a conversational style, like the kind of question you might ask a coworker. It can find answers to your questions based on information from the following tools, as well as certain types of files attached to those items:
It can also answer questions about how to do something in Procore, based on information from Procore's support site.
Copilot can answer questions about your project's RFIs, Submittals, Drawings, Specifications, and Documents. It can also answer questions about how to do something in Procore based on information from Procore's support site.
To learn more about what kinds of questions Copilot can answer, and what information it can reference in your project to provide answers, see What can I ask Copilot?
Copilot is an Artificial Intelligence (AI) tool, and its foundation is a Large Language Model (LLM). It is capable of reading a question asked in natural language, like the kind of question you might ask a coworker. It references certain information to respond using natural language, but a lot happens in between asking your question and receiving Copilot's response. Here is how Copilot works behind the scenes:
As opposed to the search engine technology we're used to, Copilot responds much better to questions asked using full sentences, instead of just a few keywords. It needs your full thought written out to interpret the intent of what you're asking, just like a person might.
Ask questions using full sentences, and include as much context and detail as you can. Simple questions about a single piece of information yield the best results.
At this time, Copilot can answer one question at a time. It can't yet answer follow up questions. It does not understand your project's history, or your role on the project. Be sure to provide Copilot with any context that would be helpful if you were to ask an intern to find the same information.
Procore Copilot, announced during Groundbreak 2023 is in open beta. The open beta is available to customers in the United States using Procore in English who license the required toolset.
Customers using Procore in English in Canada, the United Kingdom and Ireland, Australia, and New Zealand who license the required toolset are eligible to participate in a Closed Beta. Contact your Procore point of contact to request access to the closed beta.
To learn if the beta is available to your company, visit your Company Admin tool and look for the option to sign the beta agreement. Then, follow the steps in Enable Copilot to start participating.
Procore Copilot currently utilizes Microsoft’s Azure OpenAI Service. Microsoft Azure OpenAI is a listed Procore subprocessor.
No, customer data is not used to train, retrain, fine-tune, or improve Microsoft Azure OpenAI or any other Microsoft products or services, and is not provided to OpenAI. Additionally, the indexed data sent to the LLM is not persistently stored by Microsoft, only the prompts and responses are stored securely for up to 30 days by Microsoft to detect and mitigate abuse. You can learn more about how Microsoft uses prompts and data by reviewing Microsoft's Data Privacy and Security for Azure OpenAI page.
In addition to Azure OpenAI, Procore Copilot also utilizes internal AI models that help it better understand user inputs and prioritize relevant search results, improving the overall user experience. These internal models may utilize data collected from customer inputs and outputs to help improve Copilot accuracy. Procore never shares or discloses a customer’s data with another customer, and that includes customer prompts to Procore Copilot and Procore Copilot’s responses. Due to the nature of machine learning and the technology powering Procore Copilot, output may not be unique, and Procore Copilot may generate the same or similar output for third parties.
We've designed Copilot to strictly adhere to existing data permissions set in Procore. Copilot’s responses (output) are based only on data that the user can access based on these set permissions. As an example, if a user does not have access to the RFIs tool at all (based on the settings in the permissions tool), and they ask a question related to a specific RFI, then they would not get any answer based on RFIs.