For managing directors, technical directors, construction group managers and CFOs of construction firms and developers already running Business Central who want to add an AI layer where it delivers real value. Without replacing the site manager, without miracle claims, and without pulling contractual documentation out of the tenant.
Construction in Spain enters 2026 with an uneven and, above all, asymmetric level of digital maturity. Builders and developers that have rolled out Business Central with dvproject-construccion or dvproject-promocion-construccion have a solid data foundation: BC3 (Spanish construction budget exchange format) imported, daily site reports, progress certifications against the specification, subcontractors with TC1/TC2 (Spanish social- security certificates) and bond registered, change orders with their approval workflow, warranty retainage and aftersales channelled. The raw material is there.
The gap is in what is done on top of that data. Most of the analysis is still manual: the group manager opens the project dashboard to look at margins, the CFO composes the quarterly cash forecast in Excel, the aftersales lead classifies incidents by eye. There is data, but no layer yet that reads it, cross-references it and proposes action. And that layer is exactly where AI delivers real value in construction: not by replacing the site manager, but by giving them instant visibility into information that already exists in BC.
Pure developers arrive with less BC data and more weight on office tools for sales, aftersales and CPI management. The AI opportunity here runs in two directions: predictive viability analysis of the next development against financial and commercial history, and aftersales automation with incident classification and routing. Both cases first require getting the data tidy in BC with dvproject-promocion, then adding the AI layer.
The good news is that the entry cost has dropped sharply. Document Intelligence, Azure OpenAI Service and Copilot Studio are today accessible tools for a construction SMB, and the use cases with clear ROI are well-proven enough in other sectors to start without blind experimentation. What still costs is discipline: tidy data in BC, quality of the document repository in SharePoint, and the willingness to change site workflows once AI starts proposing.
Four gaps where, with BC + dvproject and the right AI layer, the ROI is clear and the timeline is reasonable. If you recognise yourself in one of them, there is a use case to start with.
A mid-sized construction firm processes thousands of invoices per month from suppliers serving several projects at once. Correct analytical posting to project, chapter and line item still depends on people who know the context. Posting errors distort the real project margin and surface late, at month-end or year-end close.
Supplier invoice automation →The site manager works on site, not in front of the ERP. Accessing a line item balance, the validity of a subcontractor's TC1 (Spanish social-security certificate) or a contract clause means calling the technical office or digging through SharePoint. Each interruption breaks the flow and pushes the manager to decide with incomplete information.
AI site manager assistant →Change orders are generated on site and are often documented late and badly. The result is change orders executed without signature, progress certifications rejected by the client and avoidable contractual disputes. AI can detect the gap between budgeted and executed work and propose the change-order draft the moment it appears.
Sector opportunity — pilot to be tailored →A delivered residential development generates owner incidents during the warranty period. Classifying the incident (concrete, joinery, MEP), assessing severity and routing to the responsible trade are done manually by the aftersales department. AI can read the owner's description, classify it and suggest the trade in seconds.
Sector opportunity — pilot to be tailored →The AI layer doesn't land in a vacuum. It sits on top of the Davisa stack you already have, or that we deploy for you before the pilot. The following flow describes the typical architecture of a builder or developer with the AI Studio live.
Three typical construction client scenarios. These are indicative examples, not published case studies with audited figures: they show how the AI Studio lands in companies with a profile similar to yours.
Problem: Finance department overwhelmed by manual posting of supplier invoices to projects. Five people spending half a day each entering invoices and looking up the correct delivery note and purchase order in BC.
Proposed solution: Smart posting pilot on dvinvoice-hub with Document Intelligence: document extraction, matching against the live purchase order and proposed analytical posting to project, chapter and line item. The invoice arrives at the validator pre-filled.
Expected result: Expected 60% reduction in capture time per invoice. Two person-days reassigned per day to control and reconciliation tasks. Conservative, indicative figures, to be confirmed with your actual data.
Problem: Each manager interrupts the technical office between 10 and 20 times a day to check line-item balances, subcontractor TC1 validity or contract clauses. The technical office runs in reactive mode and managers decide with stale or incomplete information.
Proposed solution: Site manager assistant pilot on the dvproject mobile app, with live access to BC and RAG over specifications and contracts. The manager asks in plain language from the site and gets a live answer.
Expected result: Estimated 2 to 4 hours per manager per day recovered per site. Calls to the technical office reduced by 50% to 70%. Indicative figures, to be refined in the Discovery against your actual sites.
Problem: A three-person aftersales department handles 1,500 incidents per year during the warranty period. Manual classification of the incident, lookup of the responsible trade and communication with the owner, all by hand.
Proposed solution: Incident classification pilot with AI on dvproject-promocion-construccion: the system reads the owner's description, classifies the incident, suggests the responsible trade and drafts the first response.
Expected result: Estimated 40% saving on processing time per incident. Time to first response to the owner reduced from days to hours. Indicative figures, to be validated against your actual aftersales history.
Indicative examples. Figures are confirmed in the Discovery against your actual data before committing to the pilot.
Honesty before hype. Four things AI does not solve in a builder or developer, no matter how much Discovery we run.
Three phases of the Davisa AI Studio method, adapted to the specifics of construction. Each phase with a concrete deliverable and a fixed timeline.
Field work with your technical director, two site managers and the CFO. Mapping of representative sites, assessment of live data in BC + dvproject, audit of the on-site document repository in SharePoint, and selection of the pilot case with its KPI target.
Development of the chosen use case on one or two pilot sites. Integration with BC via API, document indexing, fine-tuning with real feedback from the site team, progressive rollout. In construction, the site manager's adoption curve is the key to a successful pilot.
Rollout to the rest of the group's project portfolio, extension to the next cases in the roadmap (change orders, aftersales, AP anomalies), training of the internal team and periodic retraining of the models with post-pilot data.
Extraction, matching and analytical posting to the project for supplier invoices with Document Intelligence.
AI use caseMobile conversational assistant with live access to BC and RAG over specs and contracts.
BC extensionConstruction project management on Business Central: budget, certification, subcontractors.
BC extensionProperty development management on BC: sales, CPI, developer aftersales.
BC extensionInvoicing document hub feeding the AI pipeline with structured invoices.
SectorThe traditional sector page with the full Davisa proposition for construction.
BlogHow dvproject impacts project margin even before adding an AI layer.
BlogOdoo, SAP, Business Central and dvproject lined up for a construction firm.
HubFull catalog of use cases, sectors, method and Discovery of the AI Studio.
The reasonable minimum is Business Central in production with dvproject-construccion or dvproject-promocion-construccion live for at least twelve months, with an active budget, line items, progress certifications and subcontractors all populated. If BC is not implemented, the honest answer is to start there. AI applied to scattered spreadsheets and SharePoint folders pays back poorly in construction because site data is too dispersed. If you have plain BC without dvproject we can still talk, but the pilot scope is materially smaller.
No. AI in construction speeds up operational queries, anticipates deviations and organises documentation, but the technical responsibility on site stays human. A site manager with an AI assistant is not a junior site manager: it is the same senior with less friction to access information that already exists. The call on a change order, on a subcontractor, on cutting a line item is still made by a person. AI proposes, the person signs.
The eight-week pilot delivers one productive use case with a before-and-after metric. For invoice posting the impact is usually visible from the first week of real use (reduced typing time). For the site manager assistant, between six and eight weeks for the most-used intents to settle. For aftersales or change-order cases, results consolidate at three to four months, when there is enough volume of processed incidents for the pattern to stabilise.
No. All the AI infrastructure runs inside your Azure tenant or a managed tenant under your contractual terms. Specifications, construction contracts, signed change orders, sureties and progress certifications are indexed on-tenant with Azure AI Search. Embeddings are generated in Azure OpenAI Service deployed in your subscription. No documentation is sent to public OpenAI nor to third-party models. GDPR-compliant and ENS-compatible according to your tenant configuration.
Both, with different use cases. For builders with own or subcontracted work, the strongest cases are invoice posting, on-site assistant and change-order control. For pure developers, the weight goes to property aftersales (classifying and routing owner incidents) and predictive viability analysis of the next development based on the financial and commercial history. The Discovery decides which case to prioritise based on your activity mix.
We frame the AI layer inside your current relationship with BC and dvproject-construccion. Your usual advisor coordinates the AI Studio engagement.
Talk to the team →We start with the 5-day Discovery on your projects. Mapping of real data, technical assessment and prioritised roadmap so that in one week you know where it makes sense to invest.
Request AI Discovery →