Straight answers — including the ones where the honest answer is "no."
Who is responsible for the report?+
Your engineer of record, same as today. Report Studio produces a first draft; a licensed engineer at your firm reviews, edits, confirms the mandatory-review fields, and stamps the final document. The sign-off gate exists to make that explicit — the draft cannot be exported while mandatory fields are unconfirmed.
Does the AI compute the engineering values?+
No. The AI model computes no engineering value. It reads scanned logs and supporting documents, and it writes the site narrative from your verified data. You verify what it reads and review what it writes. Every number in the draft comes from a calculator, a live source, a document you verified, or strata your engineer reviews.
Every field is tagged with its source and its review state, so the reviewer can see which method produced a value and check it. Fields your firm designates as mandatory review block export until individually confirmed. The draft tells you which values are which — screening results are labeled as screening, presumptive values as presumptive.
Can it match our template and standard language?+
Yes. During the one-time onboarding you upload your firm's Word report template — every export renders on it, in your structure and styles. For language and engineering standards, you upload the standards documents your firm already has (code editions, bearing caps, foundation guidelines, standard phrasing); the parser stages a draft firm profile, and an admin reviews it against the source before anything goes live. A firm guideline without a written rationale is rejected.
How long does setup take?+
Onboarding is a review, not a data-entry project. You upload your existing standards documents; the parser does the extraction and presents a structured review for an admin to confirm. There's no lengthy forms exercise — the work is checking what was parsed, not re-typing your standards.
Where does our data go, and is it used for training?+
Where your data lives depends on the deployment: Civilnex-hosted SaaS, your own cloud account, or your own premises (by arrangement). In SaaS and your-own-cloud deployments, scanned log images, supporting-document text, and verified boring data go to our AI model provider, and project location goes to ASCE 7 and USDA for live lookups. Your documents aren't used to train third-party or shared models. Civilnex is built to SOC 2 controls.
Yes: on-premises (by arrangement) and your-own-cloud deployment. Live ASCE 7 and USDA lookups still need a connection.
What boring log formats work?+
Structured Excel (a blank starter workbook is downloadable from the app), PDFs with tabulated depth columns, PDFs without a ruled table (a depth ruler is fitted from the tick labels), and scanned PDFs read by a vision model. Every extracted log is shown beside your original for you to verify, and digital logs skip the AI reading step.
What can't it do yet?+
It drafts the standard geotechnical report. Analysis-level work such as slope stability, deep foundation design, and full settlement analysis is where we're headed, and pilot firms shape the order.
Does this replace our boring log software?+
No. Report Studio sits downstream of log production. Keep logging however you log — gINT, your own Excel, a field app — and Report Studio ingests the output. If a demo would only make sense as a replacement for your logging workflow, it isn't.
How is this different from ChatGPT and our Word template?+
A chatbot generates text — including the numbers. Report Studio computes every engineering value with unit-tested deterministic calculators, pulls site data live from ASCE 7 and USDA sources, tags every field with where it came from, and refuses to export a report whose mandatory-review fields haven't been individually confirmed. The Word template is the output format, not the engineering.
The fastest way to answer all of them is to see it run.