BMS Screenshot Analyser

Any screen from your building management system: trend logs, alarm history, system schematics, setpoint displays, dashboard views, AHU mimics, chiller status pages, plant-room overviews. Supported formats are PNG, JPG, and WebP. Works with screenshots from Trend, Tridium Niagara, Siemens Desigo, Schneider EcoStruxure, Honeywell Niagara, ABB Cylon, Distech, and any other BMS that outputs an image.

The AI is good at recognising common patterns (alarm trends, set-point drift, schedule anomalies, plant out of range) and at translating BMS shorthand into plain English. It is not a substitute for a competent BMS engineer's judgement — particularly on safety-critical setpoints, complex interlocks, or unfamiliar plant. Treat the output as a structured first-pass that helps you orient on what you're seeing, not as a final diagnosis.

No. Your screenshot is sent for analysis, the result is returned, and the image is then discarded — it is not stored against an account, not used to train AI models, and not visible to anyone else. For persistent storage of BMS evidence as part of an audit trail, that's the paid PM Assist platform.

The free tool analyses one screenshot per session — enough to evaluate whether it's useful on your kind of BMS. For ongoing analysis as part of an FM workflow (uploading multiple screens, attaching them to a building's documentation, sharing with the team), sign up for PM Assist's paid platform.

The AI will describe what it sees, flag anything that looks abnormal, and suggest likely interpretations (e.g. 'discharge pressure trending up over the last 24 hours suggests condenser fouling or refrigerant overcharge'). The actual diagnosis still requires an engineer who knows the specific plant — but the AI helps you orient on an unfamiliar BMS screen quickly, especially when you're not the engineer who normally looks at it.

Yes — a clear phone photo of a BMS screen works just as well as a screenshot. Frame the screen squarely, get the lighting right enough to read the values, and the AI will interpret it. This is the common case when site engineers send a photo back to the office for sense-checking before they leave site.

Yes, completely free with no signup required. The tool has reasonable per-session usage limits to keep it sustainable, but most users will not hit them in normal use. For unlimited use, attached-to-a-building storage, and team-shared analysis, see PM Assist's paid platform.

General AI tools can analyse a BMS screenshot but they aren't tuned for BMS conventions — they may misread schedule grids, misinterpret alarm priorities, or invent setpoint behaviour. This tool's prompt is built for BMS interpretation specifically, with the result structured into the parts engineers care about (overview / issues / checks / recommended next steps). It's also intended to refuse rather than invent when the image isn't clear enough.

Modern BMS installations use highly inconsistent point naming — AHU-01-RAT, ahu1_ret_air_temp, T-101-RA, etc. — depending on the integrator. The AI handles most of this from context, but if a critical point is named obscurely, add a sentence in the optional context field explaining what it is. More context produces a more accurate analysis.

Yes. The tool is fully responsive and works on phones and tablets. Site engineers commonly snap a phone photo of an alarm screen, upload it to the tool from the same phone, and get the structured analysis back in under a minute — useful as a sanity check before raising a fault ticket or calling the BMS contractor.

No — the free tool is stateless and only sees the single screenshot you upload. PM Assist's paid platform changes this: it can also see the building's commissioning records, point lists, BMS schedules, and historical alarm logs, so the analysis becomes building-specific rather than generic. That's typically the next step for FM teams who get value from the free tool.

The free tool is one-shot — useful for any site. The paid PM Assist platform is built around multi-site portfolios with per-building data separation, role-based access control, and audit logs. The application and database are hosted in Frankfurt, and document storage is in Stockholm. All customer data is stored in the EU. Held with GDPR-compliant data handling. When you ask a question, the relevant excerpts from your documents are sent to OpenAI in the United States to generate the answer. This is the only routine transfer of your content outside the EU. It is covered by OpenAI's Data Processing Addendum, which incorporates the EU Standard Contractual Clauses and the UK International Data Transfer Addendum. and never used to train third-party AI models.