Newforma puts its project records behind an AI push
Newforma’s acquisition of Toronto-based Lumi AI puts the company’s long-running project records at the center of its AI plans. Newforma intends to connect Lumi’s agentic AI to Vojo, its AI assistant, and draw on project data accumulated across more than two decades of customer use. That’s a substantial ambition: the value of an assistant depends not only on how it answers questions, but also on what information it can reach and how well that information can be connected.
The records include requests for information, submittals and project communications. Newforma plans to add natural language search and queries that can span multiple projects, giving customers a way to ask about information in ordinary language instead of relying only on file-by-file searches. The acquisition terms weren’t disclosed, and Lumi won’t be available to companies that don’t use Newforma, CEO Peter Cannone told Construction Dive. That product boundary matters: Lumi’s reach will depend on Newforma’s existing customer base and the records those customers have in its systems.
Connecting information across project tools
Cannone describes construction’s persistent data challenge as a problem of disconnected systems. Project records move among design software, construction administration platforms, document systems, email and field operations, while firms’ histories are spread across those channels. Even when information exists somewhere, finding it can take time if people have to know which system holds it, what a document is called or which project file to check.
“A project spans design software, construction administration, document systems, email and the field, and a firm’s real history spans all of it,” Cannone said. The point is broader than search speed. If a tool can connect records across a firm’s projects, it may help users see relationships that are difficult to spot when each project is treated as a separate store of information. But that depends on the records being accessible and useful to the system; a natural language interface alone can’t resolve gaps in the underlying information.
Lumi’s product lets businesses ask questions in plain language to find information and patterns in their data. Its planned role inside Vojo is to make that capability available across Newforma’s project records, so customers can search beyond a single project file. This is more significant than it looks: the proposal shifts the focus from retrieving one document to asking questions across accumulated project work. The hard part will be making the results clear enough that users can tell what information supports an answer and where it came from.
And this is the part most people overlook: cross-project search is only as useful as the scope of the records it can reach. Newforma’s description centers on its own project information, while Cannone’s account of the problem includes tools and channels beyond any one platform. The acquisition gives the company a way to extend search across its records; the announcement doesn’t establish that it will bring every system or communication channel into one searchable source.
Lumi team joins Newforma’s AI group
Newforma said the acquisition won’t lead to layoffs at Lumi. Co-founder and CTO Tudor Boiangiu has joined Newforma as vice president of AI engineering, and Lumi’s engineering team will continue working from Toronto. Keeping the team in place gives Newforma access to the people who built the technology as it works to connect that product with Vojo.
Lumi co-founder and CEO Ibrahim Ashqar said the startup set out to make data access as simple as asking a question. He said Newforma’s project information and customer base would extend the reach of Lumi’s technology. That’s a straightforward strategic fit on paper: Lumi brings a way to query data, while Newforma brings a base of project records and an existing assistant. The product work will determine whether those pieces feel like one useful tool to customers, rather than separate capabilities presented under a shared name.
If you’re working in this space, the team details are worth watching alongside the product claims. AI acquisitions can add technology and engineering capacity, but customers ultimately need the combined product to work within the routines they already use. A continuing engineering team may help preserve knowledge of Lumi’s system as integration proceeds. Still, the announcement doesn’t describe a release schedule or explain how customers will experience the connection in Vojo.
Construction software firms build around data
The deal fits a wider push by construction technology providers to acquire AI and data capabilities for their products. Procore acquired vertical AI company Datagrid in January and agreed to acquire robotics and visual intelligence firm DroneDeploy in July. Autodesk acquired Rhumbix in March, while Trimble agreed to acquire Document Crunch in April. Those deals show companies pursuing different capabilities, from AI to robotics and visual intelligence, while building out construction software offerings.
The pattern suggests that established providers see value in adding specialized technology to products customers already use. But acquisitions don’t, by themselves, show that a capability will become easier to use or deliver better answers. The practical question is whether the technology can work with the company’s existing products and data in ways customers can verify. In Newforma’s case, the central promise is access to information spread across years of project work, with queries that can span multiple projects.
The test is whether accumulated records become useful
For Newforma, the practical test will be whether Lumi can make information held across years of project work easier for customers to find and connect. Its planned cross-project search puts that accumulated record at the heart of the company’s next AI offering. That could make a long history of project information more useful in day-to-day work, especially when a question reaches across separate files or projects. A promising interface, though, won’t be enough if the underlying records are incomplete or difficult to interpret.
What this means for you is that the acquisition’s significance will be measured in ordinary searches: whether users can ask a question, find relevant project information and understand the result. Newforma has described the intended connection and the kinds of records it expects to use. It hasn’t yet detailed the product’s limits or shown how the experience will perform across different customer records. The direction is clear. The proof will be in the answers.
