[Disclosure: AgFunderNews’ parent company AgFunder is an investor in Lumi AI.]
Lumi AI—a startup enabling firms to gain actionable insights from complex enterprise data—has been acquired by project and information management company Newforma.
Under the deal, the terms of which were not disclosed, Lumi cofounder and CTO Tudor Boiangiu is joining Newforma as VP of AI engineering, while the Lumi engineering team will continue working from Toronto.
Agentic analytics
Founded in 2023 by Ibrahim Ashqar and Tudor Boiangiu, Toronto-based Lumi AI initially operated as an intelligent assistant, helping users retrieve and interpret data, generate charts and extract tailored insights that previously would have taken days to pull together by in-house teams with coding skills.
It rapidly evolved into an agentic analytics platform to proactively identify actionable insights hiding in companies’ data, with customers including supermarket chains, CPG companies, and consultancy firms.
Founded in 2003, Newforma is a software company that helps architects, engineers, and construction firms organize and manage building projects, which generate reams of data from RFIs and submittals to communication history, says the firm.
By teaming up with Lumi AI, Newforma clients will be able to extract actionable insights from the 20+ years of project data managed through Newforma’s platforms.
Moving forward, Lumi’s engine will underpin Newforma’s new AI assistant “Vojo” and will power analytics in its Konekt cloud-based project management and collaboration platform, said Newforma.
“Newforma Konekt will surface which consultant is slowing RFI turnaround, which project’s issue pattern matches one that went wrong before, and where a submittal cycle is drifting from a firm’s own baseline. Grounded in the firm’s data, with the reasoning shown.”
AI answers must be explainable, verifiable, and accurate
According to Newforma, AI that touches project decisions must meet three tests: Answers must be explainable, verifiable against the source data, and accurate enough to act on.
“Lumi built its engine around those tests. It asks clarifying questions when a prompt is vague, shows the logic behind every answer, and proactively flags data quality problems when it finds them.”
CEO Peter Cannone added: “Our customers don’t need another AI that sounds confident, they need answers they can verify.”
Lumi cofounder Ibrahim Ashqar added: “Newforma brings rich project data and a large base of firms who depend on it. Combining that with Lumi’s engine unlocks far greater value for a much wider audience.”
Solid growth vs ‘crazy’ growth
Despite triple-digit annual recurring revenue growth, a 90% pilot graduation rate and 109% net revenue retention, Lumi had struggled to secure the $10–20 million Series A round it needed to accelerate product development, Ashqar told AgFunderNews.
The decision to pursue an exit—which Lumi made at the end of May—followed a clear-eyed assessment of its ability to keep pace with rapid developments in AI without substantially more capital, said Ashqar, who is stepping back from the business following the acquisition.
“We were doing well. But they aren’t the crazy growth metrics that are expected from AI companies.”
While Lumi could have raised a smaller seed extension and continued growing, Ashqar and cofounder Tudor Boiangiu were concerned that the company would struggle to keep up with the pace of innovation, creating a “widening gap” between what it was selling and what its product could deliver, he said.
The build vs buy dilemma
For Newforma, meanwhile, acquiring Lumi offered a way to accelerate its AI ambitions without having to develop all the underlying technology in-house, said Ashqar.
Companies can develop solutions in-house, he acknowledged, but building a reliable system that connects AI models to enterprise data is harder than it looks.
“We would show them [potential clients] an article from Anthropic about how to enable self-service analytics using Claude. And what you’ll notice is they say Hey, we tried connecting Claude to our data. It was 21% accurate. We then had to build all the scaffolding [the supporting infrastructure, instructions and documentation needed to make the AI system reliable], and we were able to get it to over 95%. We left the scaffolding unmaintained for a month, and it dropped down to 65%.
“So the question is do you have the resources, the competency, the time and money to build and maintain the scaffolding, or would you rather buy off-the-shelf scaffolding that surrounds the frontier models? And that’s what Lumi has always been.”
As for the competitive set, he said, “There are other startups tackling the same problem. Some ended their journey long before us, and some are still going, although they never came up in our conversations with prospects. We were actually competing more against large [non-specialist] incumbents such as Databricks or Snowflake.”
While the deal did not deliver a blockbuster exit, Ashqar said it would safeguard employees’ jobs and maintain service for Lumi’s customers.
“It’s not like a venture-backed outcome, IPO, or multi-billion-dollar exit, but the key stakeholders have been managed well, and that’s something I’m proud of.”
