Noda, a Washington D.C.-based AI startup, launched an agentic AI platform for commercial building operations that automates the workflows of building operators and facility engineers, Commercial Observer reported. The platform operates 24/7 alongside existing teams, handling HVAC, electrical, and mechanical systems management across a deployed portfolio of 1,200 buildings totaling 350 million square feet.
Performance Claims
Across its deployed portfolio, Noda reports 80% elimination of manual workflows and a 40% reduction in the time from issue identification to completed resolution, according to Commercial Observer. Project execution rates have doubled through AI-driven prioritization, meaning a greater share of identified fixes convert into realized financial impact. Customers report 0.5-2% net operating income (NOI) growth with a 14-20x multiplier on those gains, verified against measured baselines and tied to individual assets, based on the company’s self-reported Salesforce data.
These numbers are self-reported. Independent verification of the 80% workflow elimination claim has not been published.
From 32% Accuracy to Production
CEO Kate Henningsen told Commercial Observer the technology was “only about 32 percent accurate a year ago.” The rapid improvement to production-grade reliability tracks with broader gains in agentic AI: domain-specific agents trained on structured, repeatable workflows (HVAC diagnostics, equipment scheduling, work order triage) improve faster than general-purpose assistants because the action space is narrower and outcomes are measurable.
“Although people are doing AI in buildings, we have not run across anybody who’s tackling the workflows of building engineers on site,” Henningsen said. “I think it was probably impossible to do this over a year or 18 months ago.”
The Labor Crisis Angle
Unlike most AI deployment narratives, Noda’s pitch is explicitly about filling roles that cannot be filled by humans. Henningsen cited a 15% fill rate for building operation recommendations, meaning only 15% of AI-generated maintenance and operational work orders are actually completed by the existing labor force.
“This is not a story where AI takes jobs,” Henningsen told Commercial Observer. “If you can only backfill 15 percent of your roles, there’s going to be a crisis.” She described a $150 billion annual market for building operation expenses and an acute, industry-wide shortage of facility operators.
Vertical Agents vs. Horizontal Frameworks
Noda’s approach contrasts with the horizontal agent frameworks (LangChain, CrewAI, Genkit) competing for developer mindshare. Instead of building general-purpose agent orchestration, Noda built domain-specific agents for a sector with three properties that make agentic automation tractable: deterministic workflows, high downtime costs, and repeatable decision patterns.
The template is worth watching. Capital-intensive industries with labor shortages and structured operational workflows (insurance back offices, manufacturing floor operations, healthcare facility management) share the same characteristics. Noda’s trajectory from 32% to production-grade accuracy in under 18 months suggests that vertical-specific agentic AI may reach deployment readiness faster than horizontal platforms attempting to solve everything at once.