The Intelligence Layer for Construction
Construction has digitised its tools. It has not yet digitised its intelligence. Ontolix builds the structured knowledge layer that sits beneath your drawings, schedules, specifications and platforms - so your systems, and your AI, can finally see what's connected to what.
Behind every construction project is a dense web of knowledge: which products go in which rooms, which standards apply to which finishes, which subcontractors sit on the critical path, which variations are trending toward dispute. Today that web lives in PDFs, in Procore, in finance systems, and in the heads of your best project managers.
We extract it, structure it, and connect it - creating a queryable model of your projects that makes the invisible visible. The result: you see subcontractor drift in week 3, not week 12. You see margin erosion while there's still time to act. You stop managing by exception and start managing by foresight.
Input costs rising. Productivity declining. Value capped. Margin absorbs the gap.
In construction, margin is the variable under permanent pressure. Productivity loss erodes it through the course of the build. Risk erodes it before the warning arrives. Both are consequences of an information architecture that was never built to surface either in time to act — and that is the structural gap Ontolix is built to close.
| Force | Direction |
|---|---|
| Input Costs — labour, materials, compliance, capital | ↑ Rising |
| Productivity — output per unit of input | ↓ Declining |
| Value Capture — constrained by competitive tendering | ↓ Compressed |
| Margin — the gap between value and cost | ← Absorbs the difference |
The data is structural, not cyclical. Labour productivity has grown 17% over thirty years, against 64% for the broader economy. In FY2024–25, multifactor productivity fell a further 2.8%. The industry is producing less output per dollar spent than it did a decade ago. The economic cost is $54–62 billion a year — and on a 2–5% net margin, the cost is felt project by project.
Productivity lag and margin erosion are siblings — both symptoms of the same structural dysfunction. Solve the cause and you address both.
Source: ABS · Productivity Commission · RBA · Engineers Australia, Feb 2026 · BDO 2024 · ASIC FY2025
Construction leaders already know this problem.
We spoke with project directors, commercial managers and heads of delivery across commercial and residential construction. Across project type, company size and geography, the same pressures surfaced in near-identical language.
- Senior PMs rebuilding information that already exists: submittal registers, reporting, HSEQ packs, inspection records - project by project.
- No real portfolio visibility. By the time performance issues appeared in reporting, the intervention window had closed.
- Risk read retrospectively. Subcontractor performance, variation exposure and program drift tracked narratively - the read always arrives after the fact.
What these conversations confirmed was not confusion. It was clarity about the problem - and a gap in the systems needed to act on it earlier.
"We know what the problem is. We just don't have the structure to see it early enough."
— Construction CEO, executive roundtable research
The four structural failures:
Data Sits in Silos
11.5+ hours per week rebuilding information that already exists. Operations, finance, schedules, email, SharePoint - each an island.
Assembly Replaces Analysis
High-value people reconstruct data before they can interpret it. Every report is stale before it is read.
Visibility Is Retrospective
By the time risk surfaces, the intervention window has closed. 2–5% margins turn late risk-reads into structural losses.
Program and Margin Erode Silently
Slippage compounds across trades and stages, invisibly, until it appears as a loss on the P&L.
One Structure. Every Answer.
The platform behind everything Ontolix builds is a construction ontology: a structured, continuously-updated model of how the parts of a project fit together. Extracted from the documents you already produce. Connected into a single queryable layer. Fed into every downstream system.
This is the structural fix the four failures require. A tool can only move faster inside a silo; an ontology dissolves the silos. It is why the same platform can automate today's administration, surface tomorrow's risk, and predict next quarter's program exposure - all from the same underlying structure.
What it is
A graph of every meaningful entity in your projects and the relationships between them. Not a dashboard. Not a database. A structured map of your project - where you can ask 'what depends on this?' and get an answer.
Where it comes from
Your drawings, schedules, specifications, and operational platforms - read by specialist vision-and-language models, extracted into typed entities, cross-referenced against each other and against Australian Standards and the NCC.
What it enables
Every AI system Ontolix builds queries the same ontology. That means automation, visibility, and prediction aren't separate products bolted together - they're three views of one structure. And every document you extract makes every system more accurate. The platform intelligence compounds.
The connections no project manager can hold in their head.
A senior project manager holds a detailed mental model of a handful of projects. They know which tiles are scheduled for which bathrooms, which subcontractors are double-booked, which inspections are still outstanding. It is extraordinary work - and it is the wrong layer for it to live on. When the PM rotates, the model walks out with them. Across a portfolio of thirty projects, it stops being possible for anyone.
The ontology holds these connections at portfolio scale, persistently, and queries them in seconds. Instead of waiting for the PM to notice, the system flags it. A few examples of what it surfaces today:
Floor finish on sheet A-301 (Bathroom L02-B04) does not meet NCC P3-4 slip-rating for wet areas. Standard AS 4586 applies.
J. Chen (Electrical) scheduled to Site 4 and Site 7 on 14 May. Travel time exceeds 2.5 hrs. No coverage assigned.
Steel delivery for Site C scheduled 3 days before upstream hold-point inspection (HP-042) is completed. Delivery cannot proceed.
Finish code ST02 appears on detail drawing D-114 but is missing from finishes schedule FS-03. Added to documentation defect list to review with Architect.
Examples drawn from live extraction of a 369-document Procore project: 3,004 entities, 6,592 relationships, 185 compliance edges inferred automatically.
Assistants help people. An ontology changes how the business operates.
Most construction firms have adopted AI at the assistant and workflow-automation layers. The results are real: individual productivity lifts, documents drafted faster, emails summarised. Selected tasks automated. Faster execution, same blind spots.
Assistants
ChatGPT, Copilot, embedded platform AI. Individuals work faster. The business operates the same way.
Metric: time saved per person.Workflow Automation
Document extraction, classification, task-by-task capture. Selected tasks automated. The business sees itself the same way.
Metric: tasks automated.Operational Intelligence Ontolix
AI embedded inside workflows, reading the ontology and writing back to it. Manual interpretation is removed from the critical path. Risk is surfaced, not narrated.
Metric: margin protected · program certainty · risk surfaced early.Enterprise Commercial Intelligence Ontolix
ML models trained on structured operational data. The system doesn't just show you what's happening — it tells you what's coming. Patterns across portfolios. Prediction before deviation.
Metric: portfolio forecasting accuracy · strategic decision quality.Assistants improve people. Automation improves tasks. An ontology changes how the business operates.
Intelligence is possible - but the data must be ready for it
Most AI vendor promises predictive intelligence. Very few are honest about what it requires. Prediction depends on structured, consistent, longitudinal data - and that is exactly what construction does not have by default. Most vendors skip the step of building it and sell the forecast directly. The forecast is then only as reliable as the mental model they pasted over the unstructured data.
Data as It Exists Today
Fragmented, siloed, narrative. No shared definitions. Manual extraction required for every analysis. This is where most firms operate. It is structural, not a failure of effort.
Data as Fuel Ontolix operates here
Extracted, classified, structured, cross-referenced. Common definitions applied across all projects. The ontology becomes the single data layer. This is the dependency most vendors skip.
Intelligence
ML and predictive analytics trained on the structured ontology. Program-risk signals weeks before impact. Subcontractor-performance patterns surfaced early. Portfolio forecasting. Reachable only once Zone 2 is in place - and increasingly powerful as the graph accumulates.
Firms that begin structuring now hold the longest-accumulated data when the prediction layer matures. That is a first-mover advantage that cannot be bought back.
One platform. Compounding gains.
Submittals Automation
Live40+ hours recovered per project
Specialist vision-language extraction reads construction drawings and specifications, structures every specifiable item, cross-references against Procore submittal packages, and drafts reviewed submittals automatically. Portfolio rollout underway.
Inspection Visibility
Deployed70+ live users
Real-time portfolio dashboards by project, trade and location. Inspection coverage, gaps and delivery risk derived from the ontology rather than from a template × location matrix. Embedded inside Procore.
Sales Intelligence
Deployed2% conversion lift
Pipeline visibility for leadership. Automated interpretation of sales data surfaces conversion focus areas.
HSEQ Start-Up Automation
BuiltProject start-up documentation generated automatically from the ontology. Consistent compliance output on every project. Senior leaders redeployed from administration to delivery.
Predictive Operational Intelligence
In Active BuildML models trained on the structured data the ontology accumulates. Program-risk signals, subcontractor-performance patterns, variation-exposure forecasting. Forward-looking intelligence that arrives before the reporting cycle, not after it.
Every system reads from and writes back to the same construction ontology. One model of the world. Five readings of it.
Proof before scale. Always.
AI systems are implemented only after feasibility and commercial value are confirmed.
Commercial feasibility before build. Proof before scale. Human oversight embedded - not added later.
Mark Cutfield
AI Strategist & ArchitectTo build credible AI solutions for construction, you need to understand how the business actually operates: margin exposure, delivery risk, safety obligations, quality systems, and what separates firms that protect them from firms that don't. That operational depth was developed from the inside, not observed from the outside.
Mark held commercial leadership roles inside two of Australia's leading construction firms: CMO and Business Development Officer at Shape, and Business Growth Advisor at Renascent. As business advisor to Plan Group and Chroma Group, he defined what AI must deliver to protect margin, program and quality. That operational brief became Ontolix.
He holds a certification in Artificial Intelligence, Machine Learning and Data Science from MIT, and a Bachelor of Business from AUT. At Ontolix, Mark leads problem identification, commercial case development and solution architecture - work informed by direct advisory experience inside Australian commercial construction, not frameworks.
"You can't build a credible value proposition for a construction firm without understanding operationally what creates and destroys value inside one."
Janak Mayer
Data Science LeaderTo embed AI effectively inside operational workflows, you need more than technical capability. You need experience building and leading data organisations at scale: defining strategy, shipping products, and making the case for investment in systems that change how an enterprise operates. That leadership was developed inside one of the world's largest technology companies.
Janak spent a decade at Amazon Web Services, most recently as Principal Data Scientist helping to define AI and data strategy for a global organisation of thousands. He has led multiple teams of data scientists, engineers and product managers, and built enterprise-scale AI platforms and agent systems to transform AWS's field organisation. An economist by background, earlier Janak co-founded an energy analytics consultancy advising government and utility clients across multiple continents, and led quantitative analytics at PFC Energy (now S&P Global). He holds a Master of Information and Data Science from UC Berkeley and a Master of International Relations from Johns Hopkins SAIS. At Ontolix, Janak leads AI system design, data architecture and technical product strategy.
"An AI system can't diagnose risk on a project if you haven't first turned detailed construction drawings, schedules, contracts, project plans, spreadsheets and inspection records into something a model can reason over. That structured knowledge is the foundation - everything else sits on top of it."
What CEOs See When AI Is Embedded in Operations
A productivity tool measures time saved. An operational AI system addresses the structural causes of margin erosion and productivity lag - and creates the conditions for measurable improvement across these dimensions
Time Returned to Delivery Teams
- 40+ hours recovered per project on submittals alone
- Senior PMs focused on delivery, not documentation
More Projects, Same Team
- Automation frees your best people for delivery, not admin
- In a market with chronic labour shortages, that's capacity created
Earlier Visibility of Delivery Risk
- Leadership sees risk earlier
- Portfolio-level visibility replacing project-by-project narrative
Better Commercial Forecasting
- Consistent structured data feeds ML models
- Forward-looking intelligence on margin risk and program exposure
First-Mover Advantage
- Every project generates structured data
- Firms that start now get prediction first - their data has been accumulating longest
Assistants improve people. Automation improves tasks. Operational AI protects projects. Enterprise AI protects the business.
Is this consistent with your experience?
The most useful first conversation is structured, not a vendor demonstration. A small number of questions we have found leaders return to:
Where is structural repetition consuming your highest-value project people right now?
At what point do you have real visibility into cost and program risk - and is that early enough?
How much of your reporting cycle is assembly versus analysis?
What does your board see about delivery risk across the portfolio - systematic or anecdotal?
Where would you want proof of impact before committing to scale?
Ontolix works with construction and project-based organisations moving beyond AI experimentation toward operational systems embedded in real workflows. If you are exploring where AI may improve operational performance, we welcome the opportunity to compare perspectives.