What is AI risk management in project-based industries? AI risk management in construction and other project-based industries means using AI to find early, weak signals of risk in operational data, before they show up in the monthly report. DANAOS ProjectVIEW AI is designed to scan ProjectVIEW ERP data on cost, procurement, subcontractors, progress and machinery, flag patterns that deviate from the norm, and explain them so a person can decide what to do. It does not replace the risk register or the project manager’s judgment. It helps teams see the unknown unknowns earlier. What are known knowns, known unknowns and unknown unknowns? Project risk is usually described in three groups. Each needs a different response. Category What it means Typical example How it is managed Known knowns Quantities and costs you know and can measure Concrete volume from the BoQ, agreed unit rates Estimate and budget them Known unknowns Risks you can name but not size precisely Steel price movement, weather days, ground conditions Contingency, risk register, contract clauses Unknown unknowns Risks no one has identified yet A subcontractor slowly drifting into difficulty; a cost pattern spreading across several cost codes Detected only through signals in the data, often late The first two belong in the estimate and the risk register. See what is risk management in capital projects. The third group is where AI can add something people cannot easily do by hand: watch thousands of transactions continuously and notice when something starts to move. Which weak signals can AI detect in ERP data? Unknown unknowns rarely arrive without warning. They leave small traces in transactions long before they become a claim, a dispute or an overrun. On an integrated ERP data set, ProjectVIEW AI is designed to look for signals such as: Cost. Margin erosion on specific cost codes, forecast cost at completion drifting away from budget, committed cost rising faster than progress. Procurement. Quotations above budget rates, repeated price increases from one supplier, unusual splits of orders, delivery dates slipping against WBS need dates. Subcontractors. Unapproved variations building up, delayed certifications, retention exposure, claimed progress out of line with measured progress. Progress. Activities consuming cost without matching progress, earned value indicators deteriorating across several WBS elements. Machinery. Abnormal fuel use, idle time, rising maintenance cost per operating hour. Which signal leads to which action? Weak signal Data source in ProjectVIEW ERP Recommended action (human decision) Quotation above budget rate Procurement, BoQ unit rates, cost recipes Re-quote, negotiate, or approve with a variance note Committed cost rising faster than progress Purchase orders, subcontracts, progress measurement Review the WBS element; update the forecast Margin erosion on a cost code Cost control, budget vs actual vs committed Investigate the cause; adjust estimate assumptions on future bids Unapproved variations accumulating Subcontractor and client variations, contracts Prioritise approvals; prepare evidence for claims Delayed certifications Subcontractor certifications, client certifications Escalate; check cash flow impact Abnormal fuel use or idle time Machinery management, fuel records Inspect equipment; review allocation to activities Cash flow gap emerging Receivables, payables, certifications Compare best, expected and worst-case forecasts; plan funding Why does the data structure matter? A signal is only useful if it can be traced. Because ProjectVIEW ERP ties every transaction to the BoQ ↔ WBS ↔ Cost Codes framework, a flagged anomaly points to a specific item, activity and cost code. The project manager can open the transactions behind it instead of starting an investigation from a vague alert. The same structure makes cross-project benchmarking possible. A pattern that looks normal on one project may be unusual compared with similar items on others. Is this AI predicting the future? No. It is not magic. ProjectVIEW AI finds patterns, compares them with history and expected behaviour, and estimates likelihoods such as the probability of a cost overrun. It can be wrong, and it can only see what the data contains. If progress is not recorded, or costs are booked to the wrong codes, the signals will be weaker or misleading. What AI does change is timing. ERP records what happened. AI explains why it happened and what may happen next, so teams can act while there is still room to act. Who makes the decision? A person does. ProjectVIEW AI agents follow the cycle Detect → Investigate → Explain → Recommend → Request Approval → Execute → Audit. The agent detects and explains. The responsible manager decides. Consequential actions require human approval and every step is logged. This matters for risk management in particular. A risk response often has contractual, commercial and relationship consequences that are not in the data. The AI shortens the time to see the problem. It does not take ownership of the response. How do you start? Make sure cost, commitments and progress are recorded against the same BoQ, WBS and cost code structure. Choose two or three signals that matter most to your business, for example margin erosion and delayed certifications. Agree who receives each alert and who approves the response. Review the alerts weekly for the first months and refine thresholds. Extend to further signals once the team trusts the first ones. Read more in the insight how ProjectVIEW AI exposes risk hidden in the unknown unknowns, see how ProjectVIEW ERP supports risk management, and the cost control and forecast agent. Frequently asked questions What are unknown unknowns in construction? They are risks no one has identified in advance, so they are not in the estimate or the risk register. They often appear first as small changes in cost, procurement, subcontractor or progress data. Can AI predict project risks? AI can estimate likelihoods and detect early patterns, such as the probability of a cost overrun or a subcontractor drifting from normal behaviour. It cannot predict events that leave no trace in the data, and its outputs need human review. Does ProjectVIEW AI replace the risk register? No. The risk register covers known unknowns that the team has identified. ProjectVIEW AI adds detection of signals the team has not yet identified and helps keep forecasts current. Which industries is this designed for? ProjectVIEW AI is designed for project-based companies in construction, shipbuilding and repair, marine and offshore, mining and quarrying, and project-based manufacturing. Want to see it on your own data? See how ProjectVIEW AI surfaces weak signals in cost, procurement and subcontractor data, with every decision left to your team. Book a Demo