Nine ways the agents help.
Nine named commercial real estate failure modes the agent fleet is built to catch. Each tool runs on real CRE work — a question you send, an asset you describe, a contract you want scanned. Free. No login. Pick any tool and ask.
In plain terms: AI Smart Buildings runs nine free AI agents, each built to catch one specific, named failure mode in commercial real estate — no login, no sales call. You send one question; an agent replies by email with the citations and a disclosed confidence score, usually within 48 hours.
The nine detections: (1) verify a vendor's energy-savings claim against IPMVP & ASHRAE Guideline 14; (2) catch code-triggered retrofit upgrades (IBC / IECC / CORENET X / NCC); (3) flag EVM “looks-done” reporting theater; (4) detect hybrid-policy vs. desk-count mismatch; (5) pattern-detect construction claims 30–60 days early; (6) spot vendor-KPI vs. occupant-NPS divergence; (7) compute privacy-safe occupancy data fusion (GDPR / PDPA / BIPA); (8) scan offering memoranda for hidden risk; (9) rank conflicting technical sources.
What the agents detect.
Verify M&V claims against IPMVP
Vendor claims 18% energy savings? The agent runs your case against IPMVP Option A/B/C/D and ASHRAE Guideline 14 to verify or flag.
Catch code-triggered upgrades
Touch a mechanical system in a retrofit and a chain of code triggers fires. The agent walks IBC / IECC / CORENET X / NCC against your scope.
Catch fake-looks-done EVM reports
Owner reports 80% EV but procurement is at 15%? The agent cross-checks earned value against Last Planner PPC and procurement velocity to flag EVM theater.
Policy vs. floor-plate mismatch
Policy says four days a week, headcount is 310, floor plate is 240 desks. The agent computes the peak-simultaneous demand and tells you what breaks.
Pattern-detect claims early
RFI volume doubles in three weeks alongside daily-report sentiment shifts — the agent maps the pattern against historical claim leading-indicators 30-60 days out.
Catch vendor-KPI vs. occupant-NPS divergence
Vendor scorecard 95%, tenant NPS down 12 points — the agent reads both data planes and explains the divergence.
Fuse occupancy data within privacy floor
Badge data + calendar data fused for utilization analysis — but privacy can break under GDPR/PDPA/BIPA. The agent computes the safe fusion envelope.
Scan OM for what sellers hide
Send the offering memorandum. The agent runs against a library of common omissions (deferred maintenance, near-term capex, lease tail risk, environmental).
Rank conflicting sources
Six sources, three contradict. The agent ranks them by source authority + standards anchor + numeric specificity + corroboration + contradiction check.
From your question to a cited reply.
You write the question
One paragraph with context: asset type, jurisdiction, timeline, the specific constraint or claim you want checked.
Routing picks the squad
A routing layer maps your question to the best-fit detection squad. Multi-domain questions get fanned out to multiple squads in parallel.
Agent does the research
Pulls from the squad's standards library, runs the methodology, produces a reply with the math shown, the citations linked, and the confidence floor disclosed.
You get the reply in 48h
By email. Citations included. Confidence scored. No follow-up sales call. If something is wrong, mail "correction" and we fix it publicly.
The boundary on every tool.
The agents are advisory, not transactional. They give you the analysis; you make the decision. They do not place trades, transfer money, sign contracts, or commit your firm to anything. If you forward a reply to someone with authority, you remain the one accountable for what happens next.
Pick a tool. Ask the question.
Free. 48-hour reply. Citations included. No login. No upsell. No follow-up sales call.