AI in quantity surveying: accuracy, documentation, and where you still need a quantity surveyor
AI in quantity surveying does not do the work on its own, but it changes the way the work is done: every quantity gets a source, an explanation and a sketch, and the checks that used to be skipped are done every time. I work with AI on quantity takeoffs, and I am building AiN, my bill of quantities (BOQ) software, with it. The honest answer to whether it saves time: on the first project, hardly at all. It takes patience, written working rules, an explanation for every line and a lot of computing power. What you get in the end is accuracy, order, and a document that looks good and in which every number can be proved.
What AI in quantity surveying does well
- Reading and sorting. Long specifications, drawing lists and suppliers' quotes become a list of items, units and exclusions, instead of someone reading hundreds of pages in search of one sentence.
- A calculation sheet for every item. On a TAMA 38/2 project whose quantities I calculated (TAMA 38 is Israel's national plan for strengthening existing buildings against earthquakes; track 2 demolishes and rebuilds), the bill had about 500 items, and every item had a calculation sheet with the source, the sketch and the formula. Without help, most offices do not reach that level of documentation.
- Cross-checking. Comparing the model with the schedules, the drawing with the bill, and one version with the next. That is how duplicates, swapped units and missing items get caught.
- Consistent wording. Item descriptions in the same style throughout the bill, so that a contractor understands what is being priced.
The model side has its own page: model quantity takeoff.
Where you still need a quantity surveyor
AI suggests. It does not know on its own which method of measurement was agreed on the project, where the line runs between shell and tenant fit-out, or why contractors on site price a certain item differently. Nor is it responsible for the bill. So the decisions stay with me:
- Scope and method. What is measured, in which unit, and what stays out of the bill.
- An unusual ratio. When the rebar-to-concrete ratio jumps from 140 to 178 kg per m³, someone has to open the structural drawings and ask why.
- What is missing. Missing is not zero. A value without a source is marked as missing, and does not go in as a number that looks reasonable.
- Approval. Every quantity is approved by a person who checked it. A suggestion that has not been checked is a candidate, not a fact.
Does AI in quantity surveying really save time?
Not always, and certainly not at the start. To use AI in quantity surveying properly, you have to write rules for it: how each chapter is measured, what to do with a floor that is almost typical, when to ask. You have to explain every line it got wrong, and check again. That takes a lot of patience and a lot of computing power, and it costs time on the first project of every type.
| Stage | What actually happens |
|---|---|
| First project of a new type | More work: writing rules, corrections and explanations |
| A repeat project of the same type | Less work: the rules are already written and the checks are automatic |
| Documentation and sources | A large saving: every line gets a source and an explanation without manual work |
| Version update | A saving: comparing versions flags what changed |
| Professional decisions | No saving: they stay with the quantity surveyor |
The real gain is not speed. It is an orderly bill, with proof for every number, that is easy to check, update and present to the client.
AiN as an example: what such a tool demands of the user
Building AiN taught me where AI hits a wall. Models sometimes arrive as viewing files without material layers and without volumes. One category in a model can mix structural slabs, finish layers, pavements and landscaping layers. And a quantity report that counted both a parent row and its child rows once showed 578 parking spaces instead of 289. None of these was solved with a click; each one needed a rule and a check.
That is why in AiN the AI marks every suggestion as a candidate, and a quantity goes in only after a person approves it. From the user this demands an orderly model with clear names, agreed measurement rules, and the patience to approve line after line. The details are on the AI in AiN page, and the software itself is presented on the BOQ software page.
That is the rule at every stage: an AI suggestion does not go into a quantity until a person has checked it, and a missing value stays missing. Because of this rule, the bill can be trusted.
A quantity surveyor is an engineer or a registered building technician, like Yaakov Avital, who is responsible for the bill he delivers. AI is a tool in his hands, not a replacement.
Frequently asked questions
Can AI prepare a bill of quantities on its own?
It can prepare a draft, suggest assignments and sort the material. A bill that contractors price from needs a quantity surveyor who checks every quantity and approves it.
Does it make the service cheaper?
The fee is set by the project's area and by my fee list for quantity takeoff, not by the tool. What changes is the level of documentation and checking you get.
What do I need to prepare to work this way?
The same material as always: drawings, specifications, an area schedule and a model if there is one. A model with clear names and orderly tables shortens the check.
Have a project? Tell me briefly what the project is and what already exists. When the project is a good fit, I usually reply with focused questions and a written quote.
Contact me for a quote* Created with AI for marketing, for illustration only. Not reviewed by Yaakov Avital and not professional advice. Every project must be checked on its own by a professional quantity surveyor and an engineer.