Your firm's past work, found and cited.
Twenty years of deliverables, answering the question your proposal team is asking this week, cited to the project they came from.
The firm has done it before. Nobody can find it.
For principals, engineering managers, and the project, process, and PSM engineers whose deliverable is judgment backed by records. The prior version lives in per-client folders, personal drives, and the memory of whoever ran the job.
“We already did this on another job.”
The last project's version is buried in a client folder.
New hires take months to go billable.
The gap is “where things are” and “which standard governs.”
“Which requirement governs this call?”
Code, client spec, or firm standard, it depends.
"We have done this before", with the project attached
It finds the prior projects that match the problem and stops for your read on which are genuinely comparable.
Input New scope, separator sizing, sour service
- Both used the same design basis
- The client spec was revised in 2021
Every step runs on your own documents, every output cited. See how Agentic Apps work →
Start stocked with the regs your clients live under.
Cited public regulatory knowledge, loaded before a single deliverable comes in.
Title 49 CFR pipeline safety
Parts 190–195, cited to the controlling clause.
OSHA, Title 29 CFR
5,800+ pages incl. PSM 1910.119 and 1926 construction.
PHMSA & NTSB investigations
5,000+ pages of federal incident findings, cited.
EPA petroleum standards
Refinery & terminal rules incl. SPCC 40 CFR 112.
Your firm's own reports, calcs, and the standards you hold licenses to come in alongside, so the prior deliverable surfaces cited next to the reg.
The apps we build for consulting firms
Prior Work Finder
The comparable project, found before the proposal closes.
Standards Resolver
Which clause controls, cited to the edition.
PHA / PSM Q&A
Prior studies and relief docs, traced to source.
Proposal Precedent Pack
Three comparable projects, assembled and cited.
Specific numbers, not slogans.
Straight answers.
Our staff already use AI assistants for research. What is different here?
What they are allowed to put into it. A general assistant is a reasonable research tool and a poor place for a client’s drawings, calcs and correspondence, which is why most firms have a policy saying exactly that. This runs on your own deliverables inside your tenant, keeps one client’s corpus separate from another’s, and names the page, so the precedent it returns is your firm’s prior work rather than the industry’s general opinion.
We work for multiple clients. Can we keep their documents separated?
Yes. You can stand up separate knowledge bases per client or per project, so a question against one corpus never pulls from another. Your documents stay in your own tenant and we do not train models on your data.
Does our project data stay private / on-prem?
Yes. For firms whose client agreements do not allow records off-site, knowledgeXpert runs air-gapped on your own hardware, documents never leave the building. SOC 2 Type II and ISO 27001.
What does setting this up take from our people?
Less than a migration, because there is not one. You connect the folder your deliverables already sit in, and they stay in the structure you have. The part that needs a person is deciding which practice area goes first and which sources count as approved. A pilot on one practice area can be running in days.
See cited answers running on your firm's prior work.
Thirty minutes, a slice of your own reports and calcs, and your engineers see the citations before anyone signs anything.