AI
AI that shows its work, and leaves the decisions to your people.
Ask about any unit, site or job and get an answer with the records behind it. Findings come with their evidence. Drafts wait for a person. Every call is metered against a budget you set.
The AI sits beside the page, not over it. Every Ask button offers a few questions about the record you are on, or opens a chat about it. The assistant reads your records under your own permissions, through capped lookups, and every answer ends with the units, work orders, documents and spec sections it cited.
Findings are not written by a model. Tested detectors check live points, service history, PM, warranties and material plans every day, store the records that triggered each finding, and close it when the condition clears. The AI writes a short brief per site from those findings, and a brief with a number the findings do not contain is dropped.
AI drafts, people approve. Dispatch schedules, plans, warranty claims, RFIs and takeoff marks are proposals until someone accepts them, and nothing writes to a controller. Every model call is recorded with who asked and what it cost, under a monthly budget with a warning and a hard stop.
AI drafts, people approve
Everything in AI.
Proof
Every answer ends with its receipts
The source cards are the records the answer cited, taken from its own lookups. Each one opens the unit or work order, so anyone can check the answer before acting on it.
The AI panel: an answer about AHU-012 with Sources · 4 and “Grounded in your records”
Questions
Worth asking.
Does the AI make decisions on its own?
No. It answers, detects and drafts. Schedules, plans, claims, RFIs and takeoff marks become real only when a person accepts them, and nothing in the product commands equipment.
How do we know an answer is right?
Each answer lists the records it cited, and the app reminds people to check them before acting. When the assistant cannot find a unit, it says so instead of guessing.
What can owners and reps see?
They get their own assistant, which answers only over the building, rooms or visits they were given. They never use your company chat or its tools.
How do we control AI cost?
Every call is metered at list price on the AI usage page. Admins set a monthly budget with a warning and a hard stop, and nothing is charged automatically beyond your allowance.
One record, from the spec to the service call.
We load your sites, units and documents during the pilot, so the walkthrough is on your own kind of work. The 90-day pilot is $9,000, credited to your first year if you continue.