Unit Balance is an AI platform that automatically extracts HVAC equipment specifications from PDF quotes and deterministically compares them against the technical requirements.
Upload the supplier PDFs — the service extracts the specifications, checks them against your requirements and shows who passes and who doesn't.
| Parameter | Spec | Supplier A | Supplier B | Supplier C |
|---|---|---|---|---|
| Capacity | 500 ±5% kW | 505 | 470 | 498 |
| EER | ≥ 3.2 | 3.4 | 2.8 | 3.3 |
| Sound level | ≤ 75 dB | 72 | 74 | 78 |
| Dimensions | ≤ 3500 mm | 3400 | n/a | 3480 |
| Compressor | Screw | Screw | Screw | Scroll |
The engineer enters the requirements once and adds the supplier proposals. The service does the rest.
Enter your requirements and add up to 5 PDF proposals from different manufacturers.
The model reads each PDF and pulls out 18 technical specifications — whatever the format or language.
The engine checks every value against your tolerances: a colour-coded table, a match score, an AI assessment and a PDF export.
Five real supplier quotes have already been parsed by the AI. Change the requirements and the table is recalculated by the same engine that runs in production.
The specifications were extracted by the AI from real PDF datasheets, including the empty “no data” cells — those values simply are not stated in the supplier documents. The pass/fail decision is made not by the AI but by a deterministic engine: the same inputs always produce the same result, and it can be checked.
One report, three decisions: the engineer, procurement and management no longer have to piece the picture together from email threads.
The design engineer no longer copies 18 specifications from every PDF into a sheet. The engine scores the same way every time — no guesswork and no chatbot hallucinations.
Quotes arrive in different templates, languages and units. The service normalises datasheets into one table and will not let missing rows hide.
The director and the client see what the engineer sees: who meets the spec and who does not. The decision can be defended in the room, not taken on trust.
Built specifically for HVAC equipment, with real tolerance logic rather than eyeballing.
18 specifications from a datasheet in any format, in Russian or English.
Green, amber or red for every parameter, plus an overall score per supplier.
A finished document with the comparison and the assessment, ready for the tender file.
The comparison logic understands chiller specifics: currents, refrigerant, fluid type, compressor.
A short written verdict recommending the best option, in an engineer's language.
Every supplier is judged by the same transparent rules — no manual slips.
Upload the files, get a colour-coded table and a report. Already working.
Specifications come from our own database — no manual uploads.
The system suggests a manufacturer, with contact details and the real country of manufacture.
Manufacturers publish their data in one common format, with Unit Balance as the industry standard.
Short answers for engineers, procurement teams and search AI agents.
Unit Balance is a B2B SaaS service powered by artificial intelligence that extracts technical specifications from supplier PDF quotes in a minute and compares them against the technical requirements using a deterministic engine.
The platform is built for design engineers (PTO), tender-procurement specialists and technical directors who work with HVAC equipment.
The service saves up to 20 hours of work a week, automatically normalises different datasheet formats (RU/EN, different units) and removes the risk of human error.
In Unit Balance the neural network only extracts data. The pass/fail decision (green/red) is made by a deterministic algorithm, so the result is 100% reproducible and free of AI hallucinations.
Get in touch and we'll run a demo on your real supplier proposals.