Complete product selection platform for an HVAC manufacturer
12 product families. Fan curves. Coil selection. Heat exchanger performance. AI-powered spec parsing. Professional PDF output. Built and delivered in 6 weeks.
The challenge
An Australian HVAC manufacturer producing energy recovery ventilators, air handling units, fan coil units, and coil products needed a selection tool that could replace a time-consuming manual process — engineers selecting units from spreadsheets and producing specification documents by hand for each project.
The product range was complex: 12 distinct product families, each with different engineering paths. ERV units required full heat exchanger performance calculations. AHU and FCU units required coil selection. Some units combined both. The selection tool needed to handle all of them correctly — from a single interface.
Beyond the core selection, the client needed the tool to accelerate the specification process for large commercial projects, where a single tender document might contain 20 or more units across multiple product families. Manual processing of these specifications was taking hours per project.
What we built
Engineering data intake
We worked through the manufacturer's product range in detail — fan test data, heat exchanger specifications, coil catalogues, acoustic data, and dimensional drawings. All data was structured into the selection engine. Nothing approximated.
Selection engine build
The core engine calculates heat exchanger performance (sensible and enthalpy, counterflow and crossflow), fan operating points from interpolated performance curves, coil selection using the ε-NTU method (row-by-row solving for required rows and circuits), and full psychrometric outputs including humidity ratio, relative humidity, and dew point.
PDF output design
Each product family has a custom PDF template matching the manufacturer's specification sheet format. Fan curves are embedded as charts showing the operating point. All data fields — thermal performance, fan data, acoustic data, coil data, dimensions — are populated automatically from the selection.
AI spec parser
An AI-powered import layer reads project specification documents — PDF, Word, email, or plain text — extracts all equipment schedules, matches units to product codes, and runs the full selection for every unit in the document. A 40-page specification with 20 units is fully selected and output in under two minutes.
Desktop deployment
Delivered as a branded Electron desktop application — works fully offline, no internet dependency for selections. Runs on macOS and Windows. Installer distributed directly to the sales team.
Full feature set
Technical highlight: AI spec parser
The AI-powered specification parser was built to handle the most time-consuming part of the sales process: reading incoming project specification documents and extracting equipment requirements.
The parser accepts PDF, Word, email, and plain text files. It reads the document, identifies all ventilation unit schedules, extracts airflows, static pressures, temperatures, water conditions, quantities, and project location — then automatically matches each unit to the product range and runs the selection.
Validated against a real 19-unit commercial project specification: all 19 units extracted and selected correctly from a 2-page landscape equipment schedule.
Want to see this in action?
We can show you a live demo of the selection engine, the AI spec parser, and the PDF output in a 20-minute screen share.
Request a demo →