Automated Reporting
ASHRAE 90.1 Appendix G packets. Calibration reports. ECM studies. Portfolio rollups. They used to take your engineer three days in PowerPoint. Now Roovie authors them from the common data layer. Every KPI, every chart, every narrative sentence grounded in the simulation it ran. The deliverable drops on the building and into the client portal.
▸ Traditional workflow vs. Roovie
The traditional tooling (eQUEST, EnergyPlus, Trane TRACE, IES VE) produces data. It doesn't produce a report. So the deliverable has to be re-assembled by hand, every time, for every client, in their brand, in their format. Roovie breaks that loop.
▸ From click to client: six stages, one command
▸ Why it's fast
The building model, the simulation, the context: already structured, already in one place. Every report is just a scoped query against it.
If the model doesn't support the claim, the claim doesn't ship.
A small JSON slice (not the 3.4 GB sim archive) goes to the AI.
New inputs, new deck. v1 never gets overwritten.
▸ The deliverable, generated
A real generated deck: ASHRAE 90.1 Appendix G Initial Calibration. Title slide with executive KPIs, methodology, building profile, system-by-system comparison, mandatory-provisions checklist, energy cost breakdown, HVAC baseline justification, compliance declaration, CBECS benchmark; every slide authored against the common data layer.
Every slide is constructed from a scoped data payload. The AI chooses the right visual for the data: a stacked bar for end-use, a trend for monthly cost, a comparison table for prescriptive requirements, and writes the narrative in the voice configured for your firm.
▸ The report library
A template is a schema plus a layout. Roovie ships the workhorses for energy analysis; your team adds the ones specific to your firm. Every template is a query against the same common data layer, so adding a new report type doesn't require new data pipelines.
Full performance-rating-method packet. Baseline + proposed, PCI, mandatory provisions, HVAC baseline justification, compliance declaration.
Model-vs-actual calibration showing NMBE and CV(RMSE) against 12 months of metered data, with data-quality normalizations documented.
Per-measure energy, cost, and emissions deltas: each ECM a full re-simulation, not a spreadsheet adjustment. Ranked by payback, NPV, or carbon.
N buildings in, one deck out. Executive summary, per-building calibration, targeted ECMs, portfolio-wide EUI benchmark against CBECS.
One intervention, fully characterized (surface temperature, conduction, cooling energy, emissions, ROI) scaled to a single site or a 2,810-store portfolio.
Bring your firm's deliverable format. Define the schema, map the data fields, lock the layout. Every new project generates from it: your brand, your narrative, your KPIs.
▸ Watch it happen
The generator queries the data layer. The AI writes the slides. No hand-authoring in between.
▸ Delivered, not emailed
The deck isn't a file in your downloads folder. It's attached to the building it describes, versioned against the simulation that produced it, and pushed to the client portal with a notification. Your client sees it where they expect: next to the model, next to the project, next to the prior version.
Your client doesn't have to remember which email the PDF came in. Every generated deck (PPTX, PDF, raw data) is pinned to the building it describes, inside the project it belongs to.
In-portal notifications, unread badge on the building, optional email digest. You decide the channel; Roovie fires the event the moment the render completes.
Need to send to a reviewer without a portal login? Generate a read-only link with an expiry and a password. Same deck, no account required. Revocable from the building.
▸ The audit side of it
This is the difference between a generated report and a made-up one. When a reviewer questions a KPI, your engineer can point to the exact data-layer field, the exact simulation run, and the exact query. The AI is a narrator. The data is the source of truth.
Every value on the executive KPI row is produced by a single query against the stored simulation result. Zero hand-typing. Zero transposition risk.
Before a slide enters the deck, the narrator runs a citation check. Every factual claim resolves to a field in the payload; any sentence that can't be grounded is rewritten or dropped.
▸ Re-runs are free
The second time you ship the deck, it costs nothing. Change a wall assembly, run one ECM, re-simulate; the generator re-queries the data layer, re-authors only the slides whose inputs changed, and publishes v2 next to v1 in the portal. Your client sees the diff.
Only slides whose input fields changed are re-authored. The generator walks the payload, fingerprints each section, and skips the ones that are identical to v1.
slides.changed = 8 · slides.unchanged = 17v1 and v2 both attach to the building, labeled with the simulation hash and the change summary. The old deck is never overwritten, so nothing in the client's history disappears.
v1 sim #4A1 · v2 sim #4A8 · Δ LED ECMWhen v2 is delivered to the portal, the notification carries a one-line summary of what moved (PI, EUI, savings) so the reviewer knows where to look without re-reading 25 slides.
PI 10.5% ↑ from 9.8% · $ 11,902 ↑ from 10,400