Client Overview
The client is a UK-based construction and fit-out contractor (circa £30 million turnover) operating across commercial and mixed-use developments. The business competes on speed, accuracy, and margin discipline in highly competitive bid environments, where estimating quality directly impacts win rates and profitability. To address rising bid volumes, compressed timelines, and inconsistent estimate quality, the client engaged Surtori to design and build an AI-driven cost estimation platform capable of ingesting architectural drawings and technical documentation and producing structured, defensible cost estimates for labour, materials, and fit-out activities. The objective was to reduce estimator workload, improve bid accuracy, and create a scalable estimation capability that could support growth without proportional increases in headcount.
The Challenge
The client’s estimating process was manual, time-intensive, and inconsistent. Estimators were required to:
• Interpret architectural drawings (PDF, CAD exports, schedules)
• Cross-reference specifications, finishes, and scope documents
• Manually calculate quantities and labour assumptions
• Apply cost libraries maintained across spreadsheets and legacy tools
This created several constraints:
• Long turnaround times limited bid capacity
• Estimator bias and interpretation variance affected accuracy
• Late-stage scope changes were difficult to reprice quickly
• Generic AI tools proved unsuitable due to lack of domain context, unit logic, and commercial accountability
The core challenge was not document reading — it was understanding construction intent, mapping it to build activities, and producing a cost output that estimators and commercial teams could trust.
Why a Domain-Specific Model Was Required Off-the-shelf or generic AI models were unable to meet the client’s requirements.
Generic models:
• lack understanding of construction sequencing and trade logic
• cannot reliably map drawings to labour activities and work packages
• struggle with measurement conventions, tolerances, and exclusions
• produce outputs that are difficult to audit, explain, or defend in bid reviews Surtori therefore designed a domain-specific estimation model, trained and structured around:
• construction and fit-out trade taxonomies
• labour productivity norms and regional rates
• material classifications and supplier pricing structures
• architectural conventions and drawing standards
• explicit rules for inclusions, exclusions, and scope assumptions
This ensured outputs were transparent, explainable, and commercially usable, rather than probabilistic text responses.
The Surtori Approach (Fusion Framework™)
Using Surtori’s Fusion Framework™, the engagement progressed through four phases:
Discover
• Analysed historical bids, estimates, and win/loss data
• Audited drawing formats, specification standards, and estimator workflows
• Identified high-variance cost drivers and manual bottlenecks
Outcome: Defined the scope and logic required for an AI-assisted estimation system.
Design
• Designed a domain-specific data model mapping drawings → quantities → labour → materials
• Defined AI pipelines for document ingestion, plan interpretation, and scope extraction
• Integrated the client’s cost libraries, productivity rates, and margin rules
Outcome: A controlled estimation architecture aligned to commercial reality.
Deliver
• Built an AI platform capable of ingesting architectural plans and documentation
• Automated quantity extraction and work-package classification
• Generated structured cost outputs for labour, materials, prelims, and fit-out scope
• Enabled estimator review, adjustment, and approval
Outcome: A production-ready AI estimation system embedded into bid workflows.
Embed
• Trained commercial and estimating teams on system use and governance
• Defined change-control and model-tuning processes
• Established performance tracking for estimate accuracy and time savings
Overall Outcome: Estimation capability institutionalised within the business. The Results Within the first operational period, the client achieved measurable improvements:
• 60–70% reduction in estimation time per bid
• Increased bid throughput without additional estimator headcount
• Improved cost accuracy, reducing variance between estimated and delivered costs
• Faster response to scope changes, enabling competitive repricing
• Greater consistency across bids, regardless of estimator involvement Commercial teams reported higher confidence in bid submissions, with clearer audit trails and more defensible assumptions.
Strategic Impact The AI estimation platform shifted estimation from a bottlenecked, manual activity to a scalable commercial capability. Beyond immediate efficiency gains, the system:
• created a repeatable, auditable estimation approach
• reduced key-person dependency
• improved margin protection on awarded work
• provided a foundation for future optimisation and predictive cost control
The client is now positioned to scale bid volume and complexity without compromising accuracy or profitability.
Legacy and Ongoing Impact The platform continues to evolve as new projects, suppliers, and productivity data are incorporated. Surtori remains engaged to support model refinement, governance, and performance optimisation, ensuring the system adapts as market conditions and construction methods change.
“AI doesn’t replace estimators — it removes friction, improves accuracy, and allows expertise to scale.” — Surtori, Construction & AI Delivery