PROFESSIONAL WORK · 22 September 2026
The Engineer Still Signs
Inside the first working example of Inroads' production service: a geotechnical practice's handwritten field records, returned as finished profile sheets in its own format, checked before delivery, and signed by the engineer exactly as before.
The work nobody trained for
Every professional firm carries a layer of work that is billed as professional but is not, in itself, judgement. The accountant's schedules before the review. The conveyancer's tables before the opinion. The engineer's drawings before the design. It is necessary, it is repetitive, and it is done by the most expensive people in the building because nobody else knows the house style.
Inroads Automation was built around one claim: that this layer can be taken off the professional without taking the profession off the work. The first practice to test that claim was Inroads Consulting, the geotechnical firm of Dr Brian Harrison, who has agreed to be named and who now presents on this work himself.
What the repetitive work actually is
A geotechnical investigation begins in a hole. A technician stands at a test pit and describes each layer of soil as it is exposed — its moisture, colour, consistency, structure, type and origin, in a fixed order the profession has used for decades — and writes it by hand on a field pad. Depths, samples, water, where the digging stopped and why. A recent set ran to twenty-five pits.
Back at the office, every pad becomes a profile sheet: a formal drawing of the pit in the engineer's own spreadsheet, with the description typed into the right compartment, the layers hatched with the correct symbol, depth labels centred on the contact lines, sample bars in the margin, notes numbered in the house style. Over a long career he has drawn a great many of these. Each one takes real time, and none of that time is engineering.
Learning one practice, once
The service does not arrive with its own format. It learns the practice's. Before the first pad is touched, we sit down with the engineer's own workbook, his issued reports, his technician's shorthand and his conventions for hatching, legends and notes, and we write them down as rules. That is done once per firm. It is not repeated for each job, and it is not repeated when the next set of pads arrives.
What comes out is his sheet, not ours — the same page, the same borders, the same legend, with only the content changed. On the most recent set we measured it. Comparing our rendered sheets against his own corrected file, pixel by pixel, the difference was under half a percent on every page, and every differing pixel was a change he had asked for.
What comes out is his sheet, not ours.
What the field can answer, the field answers
Handwriting raises questions. Is that a 3 or an 8. Is that word shattered or slickensided. In the old arrangement every such question travelled up to the engineer, because he was the only person who could both read the pad and remember the pit.
The route now runs the other way. The technician can photograph the pad at the pit and send it from his phone. When a word cannot be read, a cropped picture of that word comes back to him with a yes-or-no question, and he answers it standing where he wrote it. Only what the field genuinely cannot resolve — a geological call, a question of origin — reaches the engineer, and it reaches him as a short list rather than a pile of pads.
Checked before it is seen
No sheet leaves without passing seventeen checks, each written because the engineer once had to ask for it. Legend symbols clear of the frame. Contact lines that do not double up. Hatching complete to the edge of its band. Roots and gravels drawn wherever the description says they are there. Every correction he has made has become a check, which is why he does not have to make it twice.
Then the set is rendered and read again, by eye, against his last accepted work, before it goes to him. Two passes. The second exists because the first once missed something.
Every correction he has made has become a check, which is why he does not have to make it twice.
The result
On the most recent set, twenty-five sheets went from his corrected file to a rebuilt, checked and rendered set in a single morning. He opens it in the tool he already uses, reads it as he would read a junior's work, marks what he disagrees with, and sends it back. His marks become rules. The next set is closer.
The part that does not move
None of this changes who is responsible. The engineer reads the sheets. He queries what he doubts. He signs what he accepts, and his signature means what it has always meant. The service prepares; it does not decide.
That is the whole principle, and it is the reason a professional can use it. Automate the repetitive. Keep the decisions human. The engineer still signs.
The service prepares; it does not decide.
