How AI is Changing Cost Estimation in Construction - A Practicing QS Guide
The Day I Realized My Estimate Was Missing Something Important
During a tender for a large infrastructure project, I was part of an in-house team responsible for preparing the complete cost estimate. My role was quantity take-off and pricing. We had one month to submit.
Eight days into the process, while reading through the Technical Specification, I came across a requirement buried in the material testing section. The main contract said excavated material should be reused for embankment filling. Standard enough. But the specification stated that any material with a Plasticity Index greater than 15 must be disposed of at the Contractor's sole expense.
No separate pay item. No mention of this in the BOQ preambles. Just one sentence in the material testing clauses that completely changed the cost picture for the earthworks package.
That moment, finding something that would have been catastrophically missed by anyone who did not read every page of the specification, is why I believe cost estimation is one of the most technically demanding professional responsibilities in construction. And it is why the question of how AI can help deserves a careful, honest answer.
This article shares my real cost estimation methodology, the common mistakes that create financial exposure on live projects, and where AI tools genuinely fit and do not fit in the estimation process.
📚 Related Articles You May Find Useful:
My Real Cost Estimation Process : How 30 Days Actually Break Down
On that one-month infrastructure tender, here is exactly how I spent the time:
| Stage | Activity | Time Spent | Key Professional Focus |
|---|---|---|---|
| Stage 1 | Full Document Review | 8 days | Every specification section, every drawing including General Notes, geotechnical report, BOQ preambles, conditions of contract |
| Stage 2 | Quantity Take-Off | 7 days | Measurement from drawings using AutoCAD done with a colleague in parallel to cross-check quantities |
| Stage 3 | Resource Calculation | Part of pricing stage | Equipment, labour, and materials calculated per work item based on measured quantities and productivity rates |
| Stage 4 | Pricing and Unit Price Analysis | 4 days | Current market prices applied from updated price records UPAs built for each major work item |
| Stage 5 | Review and Submission | Remaining days | Full team review, checking, compilation, and formal submission |
The most time-consuming stage, without exception, was the document review. Eight days reading specifications, drawings, geotechnical reports, and contract conditions. That is where the Plasticity Index sentence was found. That is where the anti-gravity paint requirement was found. That is where most of the commercially significant risks were identified.
💼 The AI Opportunity I Missed:
Those 8 days of manual document reading if I had used AI document summarization tools at the start of that process, I could have had a structured overview of every specification section in 10 to 15 minutes. That is not an exaggeration; that is what modern AI tools can do with large document sets.
The critical point: AI summarization gives you an overview quickly. The detailed professional reading that found the PI specification sentence still required a qualified QS going through every section carefully. But having that AI overview first would have helped me prioritize where to focus my detailed reading attention, potentially compressing 8 days of document review significantly.
That time saving, applied to every tender I work on, would represent one of the most significant productivity improvements available to a practicing QS in 2026.
RICS Professional Framework New Rules of Measurement (NRM)
Construction cost estimation in professional QS practice is governed by the RICS New Rules of Measurement (NRM), the current standard for cost estimating, cost planning, and detailed measurement of building works.
| Volume | Title | Application |
|---|---|---|
| NRM 1 | Order of Cost Estimating and Cost Planning | Early stage cost estimating and elemental cost planning |
| NRM 2 | Detailed Measurement for Building Works | Detailed BOQ preparation and measurement rules |
| NRM 3 | Order of Cost Estimating for Building Maintenance Works | Maintenance cost planning and estimating |
When using AI tools for any estimation-related task, whether generating BOQ descriptions, summarizing specifications, or structuring cost reports, the applicable NRM volume must be specified explicitly in the prompt. AI will not automatically apply NRM rules unless instructed to do so.
How AI Is Currently Being Used in Cost Estimation
1. Specification Summarization
On projects with extensive specification documents, AI tools , particularly ChatGPT and Claude, can summarize large specification sections efficiently, extracting the key technical requirements relevant to cost estimation. My approach: read the full specification first to develop professional understanding, then use AI to produce a structured summary for quick reference during pricing. AI summarization complements professional reading, it does not replace it.
2. Standard BOQ Description Generation
AI can assist in generating standard, professionally worded BOQ item descriptions in accordance with the applicable method of measurement. This reduces time required to draft descriptions from scratch, particularly for repetitive or standard work items. For a detailed guide, see our article on How to Use ChatGPT for BOQ Preparation.
3. Productivity Rate Research
When productivity rate data is not available in my own data library, I use ChatGPT to research benchmark productivity ranges, always instructing it to provide real-world data with source references, not assumptions. I then verify against independent sources before applying to a live estimate.
On my current infrastructure project, I used this approach when calculating excavation productivity for a cost proposal. The Engineer subsequently challenged my productivity rate, questioning why it was low. My response was built on real site knowledge: confined area with permanent structures already in place, concurrent works requiring area-by-area excavation for safety reasons, and a 40% productivity factor applied for the rainy season. The Engineer accepted the justification.
AI provided the initial benchmark. Professional site knowledge provided the justification. That combination is how AI works most effectively in QS practice.
4. Cost Report Narrative Generation
AI tools can significantly reduce time required to produce formal cost report narratives. On a recent CTSB pavement improvement cost proposal, I used Claude AI to generate the narrative report after completing the cost estimation, providing the project data and instructing it to prepare a professional narrative including basis of estimation and resource allocation. The output was professionally structured and produced in minutes.
The Hidden Cost Traps That Destroy Estimates : Real Examples
These are not theoretical risks. They are real situations from real projects.
| Hidden Trap | Real Example | Financial Impact |
|---|---|---|
| Plasticity Index specification | Main contract said reuse excavated material. Specification said PI > 15 must be disposed at Contractor's sole expense. On large earthworks, this can mean hundreds of thousands of cubic metres of disposal cost, completely unrecoverable. | Potentially millions on large infrastructure projects |
| Drawing General Notes | Bridge project General Note specified anti-gravity paint on all pier columns to 3m height. No separate BOQ pay item. Not in specification. Only in the drawing notes. A QS who missed this would not price it in the pier construction unit rate. | Significant per pier — multiplied across all piers on a long bridge |
| Cut and fill loose factor | 1,000 m³ cut volume. 1,000 m³ fill required. Applying 1:1 ratio = zero disposal. Correct calculation: 1,000 × 1.3 loose factor = 1,300 m³ loose volume. 300 m³ surplus requiring disposal at Contractor's expense. | On 50,000 m³ earthworks: 15,000 m³ unpriced disposal |
| Structural backfill included in excavation rate | Specification stated excavation pay item included "necessary disposal of unsuitable materials and structural backfilling." No separate structural backfill pay item. Must be included in excavation UPA; missing this means unrecoverable backfill cost. | Significant on any project with substantial structural excavation |
| Material standard compliance | Specification referenced AASHTO material standards for aggregates. Local market materials did not fully comply with the imported materials required. Significant cost difference between local and imported supply. | Can be substantial on material-intensive packages |
Common Estimation Mistakes and How AI Can Help
| Common Mistake | Professional Impact | How AI Can Help |
|---|---|---|
| Not reading Drawing General Notes | Missing cost-significant requirements hidden in drawing notes | AI can summarize drawing notes and flag cost-significant requirements for professional review |
| Missing material standard requirements | Significant cost difference between locally compliant and imported materials | AI can scan specifications for material standard references, AASHTO, JIS, ASTM, BS |
| Ignoring cut/fill loose factor | Massive underestimation of disposal volumes on earthworks projects | AI can generate the correct calculation framework professional applies to project-specific factors |
| Using outdated market prices | Significant variance between estimate and actual project cost | AI can assist in researching current market benchmarks, always verify with local market data |
| Excel formula errors | Calculation errors in cost spreadsheets common and potentially costly | AI can review formula structures and identify logical errors, reducing spreadsheet risk |
What AI Will Never Replace in Cost Estimation
1. Commercial Negotiation
Cost estimation leads directly to commercial negotiations. On my current project, I defended a 40% productivity factor for excavation works to the Engineer, explaining confined access, concurrent works, and rainy season constraints. That real-time commercial defense requires human professional judgment, direct site knowledge, and interpersonal skill. AI cannot sit in that meeting.
2. Site-Specific Judgment
The correct productivity rate for excavation in a congested infrastructure project during rainy season is not a number you find in a data library. It is a judgment call based on direct site experience. AI can give you a benchmark. Only you can assess whether that benchmark applies to your specific site conditions.
3. People and Relationship Management
Construction projects involve large numbers of people across multiple organizations. Managing the commercial relationships with Engineers, subcontractors, suppliers, and clients requires human professional capability. The construction industry is fundamentally a people industry, and that will not change.
Professional Cost Estimation Software : What the Industry Uses
Dedicated QS Estimating Software
| Software | Primary Application | Common Markets |
|---|---|---|
| CostX | Quantity take-off from 2D and BIM drawings, BOQ preparation, estimating | UK, Australia, Middle East, Asia personally used, excellent for building projects |
| Candy CCS | Estimating, project control, planning and cost management | Africa, Middle East, South Asia |
| Buildsoft | Estimating, tendering and cost planning | Australia, Southeast Asia |
| Cubit | Quantity take-off and estimating | Australia, New Zealand |
| STACK | Cloud-based estimating and take-off | Global cloud platform |
| Bluebeam Revu | PDF-based quantity take-off and markup | UK, USA, Australia |
Construction Management Software with Cost Features
| Software | Primary Application | Cost Estimation Capability |
|---|---|---|
| Primavera P6 | Construction scheduling and programme management | Cost loading and budget tracking |
| Autodesk Construction Cloud | BIM-based project management | Model-based quantity extraction and cost management |
| Procore | Construction project management | Budget management and cost control |
| Microsoft Excel | General spreadsheet application | Widely used formula error risk requires careful management. AI can assist in reviewing formula structures. |
Important note: Software selection should be based on project type, contract requirements, company standards, and professional judgment — not marketing. Each platform has specific strengths that must be assessed against your specific project needs.
My Honest Advice Starting With AI in Cost Estimation
Start with specification summarization. This is the lowest-risk, highest-immediate-value application. Read the full specification yourself first, then use AI to produce a structured summary. This gives you overview speed without sacrificing the professional depth that finds the Plasticity Index sentences.
Use AI for standard BOQ description generation. Significant time saving, low risk, provided every description is checked against the applicable NRM measurement rules before finalizing.
Never delegate full cost calculation to AI. All calculations must be professionally reviewed and verified. The professional remains accountable for every number. AI does not share that accountability.
Check everything yourself. Every piece of AI output quantity, rate, description, or total must be independently verified before professional use. This is not optional. This is professional practice.
Conclusion
The Plasticity Index sentence was found because I read every page of that specification. The anti-gravity paint requirement was found because I read every Drawing General Note. The cut/fill loose factor was applied correctly because I understood how soil volume changes after excavation.
None of those insights came from AI. They came from professional knowledge, systematic document review, and the discipline to read everything, not just the obvious sections.
AI tools can make the document review process faster. They can help with BOQ descriptions, productivity rate research, and cost report narratives. These are genuine time savings that allow a QS to redirect professional attention to the commercially critical work.
But the commercially critical work finding the hidden obligations, making the site-specific judgment calls, defending your rates in a meeting with the Engineer that remains yours. It always will.
Use AI to work faster. Use your professional knowledge to work better. On a construction project, both matter.
💬 Found This Useful?
If this article helped you — please consider:
- ⭐ Sharing it on LinkedIn — the hidden cost traps section could save a colleague significant money
- 💬 Leaving a comment below — what is the most dangerous estimation mistake you have seen or made?
- 📖 Reading our related articles — links below
Have you ever found a hidden cost obligation that others missed? Share your experience in the comments!
About The Author
A practicing Quantity Surveyor with extensive experience in large-scale infrastructure projects under FIDIC conditions.
Qualifications:
- BSc (Hons) Quantity Surveying
- MCIOB — Chartered Institute of Building
- MQSI — Quantity Surveyor Institute Member
This blog shares real professional experience from live construction projects — not theoretical content.
📚 Related Articles You May Find Useful:
Frequently Asked Questions
Can AI prepare a complete construction cost estimate?
What is the RICS guidance for construction cost estimation?
What is the most dangerous hidden cost trap in construction estimation?
What is the cut/fill loose factor and why does it matter?
Can AI help with quantity take-off from drawings?
Is Excel still reliable for construction cost estimation?
Comments
Post a Comment