PENGA
One learning agent for hydronic retrofit intelligence.
PENGA is being shaped as a control layer for legacy hydronic buildings: a trained agent that integrates thermal physics, tenant comfort constraints, Alberta energy market timing, retrofit safety, and operator trust into one operating strategy.
01 / The Problem
The problem is integrated.
Thermal inertia
Concrete, water, envelopes, weather, and boiler cycling create a slow physical system that outdoor reset cannot reason about.
Tenant comfort
Comfort is not just a fixed air temperature. PENGA's product direction incorporates ASHRAE/PMV-informed comfort boundaries and tenant experience risk.
Energy markets
AESO price volatility turns equipment timing into an operating decision as buildings add controllable electrical loads.
Retrofit friction
Legacy buildings often lack modern BAS access. The wedge has to respect existing equipment, certified operators, safety controls, and budget reality.
02 / The Agent
Not a dashboard. Not a thermostat.
The product thesis is a trained operating policy that learns when to heat, coast, protect comfort, and preserve equipment. The prototype stack combines an RC thermal model, hydronic loop simulation, SAC-Lagrange training, comfort-constrained rewards, AESO/weather forecasts, an OAT sensor intercept deployment path, and a watchdog fail-safe that defaults toward heat on fault.
03 / Business Wedge
The first retrofit earns the data for a broader control platform.
Gas optimization
Start with the simplest retrofit surface: better operating intelligence for existing central hydronic boilers.
RTU optimization
Extend the same control discipline to rooftop units where short cycling, demand peaks, and comfort complaints are visible operating costs.
BTM Li-ion retrofits
Use building load data to identify where behind-the-meter batteries can reduce peaks, shift consumption, and support resilient operations.
Fleet response
Aggregate hydronic systems, RTUs, and BTM storage into a fleet-managed virtual power plant once enough sites are instrumented and trusted.
The first sale is not the whole company. It is the first trusted operating surface: measured site data turns a retrofit controller into a calibrated understanding of how each building stores heat, consumes power, and can participate in managed load programs.
04 / Current State
What exists now.
Prototype revision
Simulation, training, benchmarks, controller interfaces, safety wrappers, and edge deployment research are already represented in the working codebase.
Benchmark evidence
Prototype evaluations on historical weather and AESO data show meaningful savings potential with explicit comfort constraint tracking.
What comes next
The next phase is practical validation: the right introductions, the right buildings, and the right ecosystem support to move from prototype revision into live inquiry and pilot shaping.
Aligned Interest
We are looking for the right conversations, not just capital.
GlacialRift is building from a working prototype toward real-world validation. The best next step is a flywheel of aligned inquiry: local innovation support, technical resources, and introductions to buildings that genuinely fit the niche.
- Innovation ecosystem contacts with grants, pilot programs, technical resources, or local support.
- Commercial real estate owners, operators, and property teams with older hydronic buildings that fit the profile.
- Partners who can connect PENGA to the right buildings, operators, or early deployment environments.
••••@••••••••••••.••
Reveal to copy the address.