The most expensive questions in surveying are the ones a firm can already answer and doesn't ask, because reaching the answer takes an afternoon the decision doesn't allow. Have we worked near here? Did we set control next door? Is there a utility we already hit? Theo, FieldIntel's assistant, turns that afternoon into a plain-language question answered in seconds -- from the firm's own plats, field notes, and deeds. And it is built for a profession that cannot act on a guess: every answer is grounded in a specific record and cites it -- the job number, the field note, the page, the date -- and when Theo does not have it, it says so instead of inventing one. Ask before you break ground, and you show up to the bid, the site, and the dispute already knowing what you knew.
The question you already know the answer to
Start a new project – a boundary, a topo, a route – and there is a question sitting under it that would change how you approach the whole thing: what does this firm already know about this ground?
Over decades, the odds are you know something. A job next door. Control you set two ridgelines over. A boundary your firm resolved in a subdivision that shares this parcel’s line. A benchmark, a monument, a utility conflict somebody logged. That knowledge, if you had it in hand, would sharpen the bid, shorten the mobilization, and steer the crew around the surprise.
But you usually do not have it in hand. It is spread across the archive and a few long-tenured people’s memories, and reaching it means an afternoon in the filing cabinet that the schedule does not include. So the question does not get asked. The decision just gets made without it – not because the answer is unknown to the firm, but because it is unreachable in the window the decision lives in.
Three moments it changes the answer
That window shows up at three specific moments, and at each one, knowing versus not knowing is real money.
At the bid. You are pricing a topo on a parcel, and the honest input is whether you have usable prior work – control you already set, a corner you already found. Ask Theo “any prior work within a half mile of this parcel?” and it comes back with the jobs: Meridian Parkway Phase II, 2019, three control points within 500 feet, a found iron pin at the northeast corner. Now you price the efficiency you already earned instead of pricing blind and either losing on number or giving the margin away.
At the mobilization. Before the crew rolls, ask what is already established out there. Knowing, they drive to productive work; not knowing, they spend the first morning re-establishing a benchmark the firm set six years ago. The difference is a question asked the night before instead of a discovery made on the clock.
In the field. The party chief is standing at an intersection and asks Theo “any utility conflicts near here?” – and gets back the note from a 2017 job: eight-inch water main, 4.2 feet south of the centerline at this crossing. That is the difference between a routine set and a strike, a change order, and an afternoon lost. The firm knew. The question was whether the person on the point could reach what the firm knew, right then.
What “grounded” means – and why it is the whole point
Here is the part that separates an assistant a surveyor can actually use from one that is a liability waiting to happen.
Surveyors are right to distrust AI. An assistant that will confidently produce a coordinate it never computed, or invent a plausible-sounding prior job, is not just unhelpful in this profession – it is dangerous, because someone might stamp work that leaned on it. A chatbot that is usually right and occasionally, invisibly, fabricates is the worst possible tool to put near a legal instrument.
Theo is built the opposite way. It does not generate a plausible answer; it retrieves the answer from the firm’s records and shows its work. Every figure it states traces to a specific source you can open – the job number, the field note, the page, the date it was scanned. “BM-44, 842.67 feet NAVD88, set 2019, Job #4412, last verified 2024” is not Theo’s opinion; it is a pointer to a record, with the record attached. And when Theo does not have the answer, it says so plainly rather than filling the silence with something that sounds right.
That is not a nice-to-have feature bolted onto an AI. It is the requirement. In work that carries a professional seal, the only AI answer worth acting on is one you can trace back to the record it came from and verify in ten seconds. An assistant that cites its sources is a tool. One that cannot is a party trick you keep well away from anything you stamp.
Ask before you break ground
None of this replaces the surveyor’s judgment. It feeds it – earlier, and with evidence attached. Theo does not decide what a prior job means for this boundary or whether that old control is still good; it puts the prior job and the old control in front of the person who will decide, in the moment they are deciding, sourced so they can trust it.
The firm that asks first shows up to the bid, the site, and the eventual dispute already knowing what it knew. The firm that does not rediscovers it – re-prices work it could have priced sharp, remobilizes to find what it already set, or hits the exact thing a crew warned about eight years ago. The knowledge was there the whole time. The only variable is whether it is reachable in the window that matters, grounded well enough to act on.
The archive was always the firm’s memory. Theo is what makes it answer when you ask – before you break ground, not after you wish you had.
See what your own archive can answer
That is the FieldIntel design-partner program: we profile your firm's archive on your own data, structure decades of plats, field notes, and deeds, and let you ask Theo the questions you would normally spend an afternoon chasing -- grounded in your records, sourced so you can trust the answer. Comped onboarding, half price the first year for design partners.
Schedule a 30-minute callOr browse the plans → stratalogic.io/purchase