Your symbols, your assembly rules, your conventions, written down once and answerable with the page they came from.
Each week ends in something you can hold: your standards written down, your record answerable, a document you keep.
We ingest ten to twenty past project sets, your standards, your assumption library and your output formats. We also record your current baseline: how many hours a set takes your team today, measured from your own time records. That baseline is the number everything is judged against.
Your team puts the record to work on the one live project you picked. They ask it what the firm decided about an assembly, when, and who signed off. Every answer opens to the page it came from, so a person checks it in seconds.
Your symbol library, your assembly rules and your conventions, written down in one place. You keep it, you can export it, and it stays yours if nothing else happens.
We add the next workflow in the order you tell us hurts most. Energy-code calculations and U-value schedules, zoning review, then quantity production if you want it.
Week one is the whole engagement in miniature. If your standards or your baseline are wrong, everything downstream inherits the error.
Whatever you have, in the format you already keep it in. Native files if they exist, PDFs if they do not. Older sets are still useful, because the point is how your firm has answered these questions before.
Sheet naming, layer conventions, room numbering, the abbreviations your team actually uses. If it lives in one person's head, we write it down during week one.
The schedule format you hand to a client or a GC. The engine produces that shape from the start, so nobody on your team has to rebuild the output by hand.
The defaults your team applies without discussing them. What counts as a door type. When a wall length is measured to the face and when to the centre. What gets rounded and how.
Taken from your own time records, on sets you have already produced. We do not estimate this number for you and we do not accept a guess, because it is the number the whole engagement is judged against.
You choose it, and weeks two and three run on that one type. Depth on a single type is what makes the engine hold up when we add the second.
Eight things. All of them yours to keep.
Your symbol library, your assembly rules and the conventions nobody wrote down. One document, exportable, and it stays yours whatever you decide next.
Ask what your firm decided about an assembly, when, and who approved it. Every answer opens to the email, markup or sheet it came from.
Doors, windows, rooms, fixtures, wall lengths and areas, when you ask for them. Every line stays connected to the drawing, sheet and location it came from.
Your naming, your column order, your assumptions. The file opens looking like your firm produced it, because the format came from your own templates.
What the drawing set does not answer, listed out. Unanswered questions in a set are what bring it back from the GC, so they are worth seeing before it ships.
Your baseline hours from week one set against where they land, taken from your own time records and signed by you at the end of week four. If the number does not move, it says so.
Documented and yours. If you walk away it goes with you, and we do not train on your firm's data for anyone else.
Review is a named responsibility on the engagement, following the AIA's guidance on responsible AI use in architecture and design firms.
If you walk away, the operating memory goes with you. We do not train on your firm's data for anyone else.
The memory is built from your past sets, your standards, your templates and your assumption library, and it improves with every set you send. That is what makes the output match how your firm actually works. It is documented so a new hire can read it, and exportable so it stays available to you whatever happens to the contract.
You already have options, and each one stops somewhere specific. Here is where.
Nobody on your team learns an interface or types commands. Document search returns the passage closest to your question, and it stops there. Who approved an assembly, against which clause, and when, is a path across four documents, an email, a markup and a date. We work on top of the tools you already own.
An outside firm does the work well and never learns yours. Standards get re-explained every engagement, and the accumulated knowledge leaves when the contract ends. The published figure for a single request for information is about eight hours and $1,080, at 796 of them on an average project, and every one of those is a question somebody had to re-answer.
A senior hire costs a salary, takes months to learn how your office actually does things, and the context leaves when the person does. The written standard stays after the person is gone.
Earthwork and cut-fill quantities, geotechnical analysis, soil and rock-bed conditions and existing-condition assessment are out of scope. Those quantities come from the ground. We serve the phases a drawing set can answer, and we say so before you ask.
Your standards are the entry point. The order after that is yours, because you know which workflow is costing your senior team the most time.
The memory already holds your standards, so a second and third project type is an extension of work that is done.
Compliance-path calculations organised for professional review, and coordinated fenestration U-factor and insulation R-value schedules, including the conflicts between what is drawn and what is specified.
Setbacks, height limits, lot coverage, floor area and parking compared against current project information, with a preliminary buildable-envelope diagram.
A structured technical assessment drafted from the plans: project description, area and program tables, assumption register, and a risk and missing-information section.
A recurring workflow specific to your firm, mapped first and then built around your templates, terminology, approval structure and existing tools.
That is a delivery limit. Week one is hands-on and the engine is built around one firm's standards, so ten is what we can do properly in a quarter.
If yours is not here, ask it on the call. We would rather disqualify ourselves on the first call than in month three.
Ten to twenty past project sets, your drawing and naming standards, your output templates and your assumption library. We also need forty-five minutes with whoever knows how your office actually does things. You pick the one live project weeks two and three run on.
A qualified professional, named on the engagement. Every result shows which documents were used, which drawing page supported it, which rule was applied, what assumptions were made and what could not be found, so a reviewer can verify a set in about five minutes. This follows the American Institute of Architects' guidance that architects remain accountable for all work products, including work produced with AI assistance.
You keep it anyway, and thin is itself an answer. Most firms find the symbol library comes out fast and the conventions are where the room goes quiet. If your standards turn out to be genuinely written down already, we say so, and the four weeks were ours to lose.
No. The operating memory is yours, documented and exportable. If you walk away it goes with you, and it is not used to train anything for anyone else.
Design and construction firms where the answer to a question about a past project lives in one person's head. Architecture, structural and civil engineering, general contractor preconstruction, and the electrical, plumbing, roofing and glazing trades. The test is whether your standards are written down anywhere a new hire can read them.
Week one is hands-on, and the engine is built around one firm's standards. Ten is what we can deliver at that depth in a quarter. It is a delivery limit.
The Standards Gap Test asks 40 questions about your symbol library, your assembly rules, and the conventions nobody wrote down. You answer from memory in 15 minutes, then you count how many answers only one person can give.
Forty-five minutes. We map your symbol library, your assembly rules and the conventions nobody wrote down. You keep the map whether or not anything else happens, and if we cannot help we say so on the call.
