AI for PropTech
PassiveLogic logo

PassiveLogic

Autonomous building control platform using physics-based digital twins and differentiable AI to run and optimize building systems at the edge. Its Hive controller makes millisecond-level decisions on-site without a cloud connection.

Best for: Building owners and operators retrofitting HVAC and mechanical systems for autonomous, energy-optimized control

Last reviewed 6/2/26

Category: Maintenance & Operations Autonomous building control
Tags: autonomous-buildingsdigital-twinbuilding-automationedge-aihvacenergy
Integrations
BACnet/IP, BACnet/IPv6, BACnet MS/TP, Modbus TCP, Modbus RTU, KNX, Zigbee, BLE Mesh, 1-Wire
Property types
Commercial, Office, Institutional, Industrial
Geography served
Global (HQ US; offices in Amsterdam and New Delhi)
AI

How PassiveLogic uses AI[1][2]

Autonomous building control runs on the edge through PassiveLogic's Hive platform, which uses physics-based 'Quantum' digital twins and a differentiable AI compiler to optimize HVAC and energy in real time.

  • Quantum digital twins model a building's zones, weather, and mechanical physics
  • Hive runs millions of simulations in seconds to pick optimal control paths for the hours ahead
  • Edge hardware (12-core processor plus on-board NVIDIA GPU) keeps decisions local, no cloud required
  • Differentiable Swift compiler trains models at the compiler level; PassiveLogic reports 322x faster than TensorFlow and 238x faster than PyTorch on a building thermal model

AI type: Physics-informed AI, Differentiable Programming, Digital Twins, Predictive Control, Edge AI

API: unknown MCP: unknown

Key numbers[2][3][5]

  • $74M Series C in Sept 2025, total funding over $125M
  • partnered with NVIDIA to retrofit NVIDIA campus buildings
  • reports differentiable Swift compiler is 322x faster than TensorFlow and 238x faster than PyTorch on a building thermal model
  • Hive runs fully on the edge with no cloud connection

Credibility[3][4][5]

Founders
Troy Harvey (Co-founder & CEO, formerly CEO of engineering firm Heliocentric); Jeremy Fillingim (Co-founder, formerly partner at Mote Systems)
Customers
NVIDIA (partnered with PassiveLogic to retrofit campus buildings); building owners, operators, architects, engineers, contractors, and utility partners
Investors
noa (lead, Series C); Prologis Ventures; Johnson Controls; PSP Growth; Addition; NVentures (NVIDIA); Keyframe Capital; Brookfield

Founded

2016

Headquarters

Salt Lake City, UT

Stage

Series C

Employees

100-200

Related companies

Noda logo

Noda

AI-powered building orchestration platform for commercial real estate that optimizes energy consumption, reduces costs, and lowers carbon emissions across portfolios.

Agentic AI ingests and normalizes data from BMS, meters, and sensors to build digital twins of building systems, then ranks energy optimization opportunities by economic impact and automates the control changes that capture them.

Maintenance & Operations Smart Building energy-optimizationsmart-buildinghvac
View profile
Virtual Facility logo

Virtual Facility

Centralizes building system alarms and uses AI to filter non-actionable noise, prioritize events by safety and uptime impact, and auto-generate CMMS work orders.

Machine learning triages building alarms, prioritizes the events that affect safety, uptime, or comfort, and converts validated alarms into work orders inside a facility's existing CMMS.

Maintenance & Operations Smart Building alarm-managementalarm-triagecmms-integration
View profile
HappyCo logo

HappyCo

AI-powered maintenance operations platform for multifamily that prioritizes and routes work orders, transcribes voice completion notes, and triages after-hours calls without human intervention.

JoyAI, HappyCo's custom-trained AI companion, runs centralized maintenance for multifamily operators: it enriches and categorizes work orders, matches technicians by skills and location, and powers 24/7 resident communications.

Maintenance & Operations Work Orders multifamilymaintenance-operationsinspections
View profile