Newforma has launched Vojo, an AI assistant built on an agent-driven framework, and its choice of what to automate first is revealing. Not design. Not estimating. Not site monitoring. Vojo’s flagship capabilities target the administrative connective tissue of a project — finding information, reviewing submittals, and filing email — the unglamorous work that quietly consumes a staggering share of every architect’s and project engineer’s week.
Announced at the company’s Newforma World 2026 event in May, Vojo is a deliberate bet that the highest-ROI place to put AI in construction is not the work people find exciting, but the work they find tedious.
What Vojo Does
Vojo’s capabilities cluster around four jobs, all of them familiar to anyone who has run a project:
- Natural-language search across project data, so a team member can ask a plain-English question and get an answer drawn from the project record rather than hunting through folders and email threads.
- AI-powered analysis of BIM models and documents, extending search into the structured and semi-structured content that holds most of a project’s real information.
- AI-assisted submittal review, automating one of the most repetitive and deadline-sensitive workflows in construction administration.
- Smart Email Filing, which automatically associates incoming email with the right project and recommends where to file it — flagging RFIs, submittals and action items along the way.
That last feature deserves the attention. Email filing sounds trivial; it is anything but. In the architecture-engineering world Newforma has served for two decades, the project record is the email — the decisions, approvals, change directives and liability-defining correspondence that determine who is responsible when something goes wrong years later. The single biggest threat to that record is not malice, it is entropy: messages that never get filed, get filed in the wrong place, or get filed in one person’s inbox and nowhere else. An incomplete or fragmented project record is a legal and operational time bomb. Automating the capture of that record is exactly the kind of boring, high-stakes problem that AI is genuinely good at and humans reliably neglect.
”Lower Risk, Save Time, or Unlock Scale”
Newforma’s chief product officer, Carl Veillette, offered a tidy framing for the strategy: “Everything we build must either lower risk, save time or unlock scale.” It is a useful lens for reading Vojo, because each feature maps cleanly onto one of those three. Smart Email Filing lowers risk by keeping the project record complete. Submittal review and natural-language search save time on tasks measured in hours per week per person. And the agent framework underneath is the bid to unlock scale — to let a fixed team manage more projects without a proportional increase in administrative drag.
It is also a quietly pointed contrast with much of the construction-AI market, where the pitch is often a dramatic transformation of how work gets done. Vojo’s pitch is the opposite: keep doing the work the way you do it, and let an assistant absorb the overhead. For Newforma’s core customers — AEC firms whose competitive constraint is professional labour, not technology — that framing is likely to resonate more than a promise to reinvent their workflow.
The Open-Ecosystem Play
Alongside Vojo, Newforma introduced an “open ecosystem” strategy that may matter as much as the assistant itself. The company expanded its integrations to include connectors for Bluebeam, Egnyte, Microsoft Teams and Autodesk Build, and moved to a simplified, modular licensing structure aligned to specific user roles.
This is a sharp strategic instinct. The reality of an AEC firm’s software stack is that it is heterogeneous and will stay that way — a Newforma project lives partly in Autodesk’s tools, partly in Bluebeam, partly in Teams, partly in cloud storage. An AI assistant that only knows what is inside Newforma’s own walls is half-blind. By positioning Vojo on top of an open set of connectors, Newforma is betting that the assistant’s value comes from reasoning across the whole stack, not just its own slice of it.
It is the same insight driving the platform giants’ consolidation — that the prize is the connected flow of information across tools — but pursued from the opposite direction. Where Autodesk is buying its way to an end-to-end stack, Newforma is trying to be the neutral layer that sits across stacks it does not own. Both are responses to the same problem: a project’s information is fragmented, and whoever reunifies it captures the value.
Where It Fits
Vojo enters a crowded field of construction “copilots.” Procore has rolled out its own construction-focused AI agents; Autodesk, Trimble and Nemetschek all have AI layers in flight. What distinguishes Newforma’s entry is its narrowness and its heritage. Newforma did not start as a construction-management suite trying to bolt on email and document handling; it started as the email and document handling. Project Information Management is the company’s native discipline, and Vojo is AI applied to the problem Newforma has always owned.
That focus is a strength and a limit. The strength is credibility: an AI assistant for project information from the company that defined the category is an easier sell than the same feature from a platform stretching into unfamiliar territory. The limit is ambition: Vojo is explicitly an assistant that makes existing administrative work faster, not a tool that changes what the work is. For Newforma’s customers, that is almost certainly the right altitude — the gains are real, the disruption is low, and the boring problem it solves is one they actually have.
The interesting question Vojo poses to the rest of the market is whether the industry has been over-indexing on the spectacular and under-indexing on the mundane. Autonomous excavators and generative design make better headlines. But the hours quietly lost to filing, searching and reviewing add up to a tax every firm pays every day. Newforma is betting that collecting that tax back is a product worth shipping — and it is probably right.
