How Construction Plan Verification Actually Works in 2026

Construction plan verification is the process of checking project documents — plans, technical studies, specifications — against every applicable building code, owner requirement and industry standard before construction begins. In 2026 it happens through three different models, often in combination: the statutory contrôle technique regime in France, the Gateway approval regime under the UK's Building Safety Act, and the local plan-check process under the IBC in the United States. All three have been under pressure for the same reason — regulations have grown faster than the workforce verifying them — and all three are now being augmented by AI to restore full coverage at modern project speeds.

What plan verification is, in plain terms

Plan verification is the work of reading a project's documents and confirming, line by line, that what is drawn complies with what is required. The work cuts across structural safety, fire safety, accessibility, energy performance, environmental classifications, owner-specific standards (such as a hotel brand's room programme), and any local urbanism rules attached to the site. The output is a structured report identifying every non-conformity, locating it on the plan, ranking it by impact, and recommending a remediation.

This is not the same as inspection. Inspection happens on site, on what is built. Verification happens before construction, on what is proposed to be built. The economics of the two are very different: an error caught in verification costs minutes of rework on a drawing; the same error caught in inspection costs weeks of demolition and reconstruction.

Who does it, by jurisdiction

The legal architecture of plan verification differs by country, but the function is broadly the same in each. What changes is who carries the obligation, what is mandatory, and what gets cross-checked against what.

France: the contrôle technique regime

The French model was set up by the Loi n° 78-12 du 4 janvier 1978, known as the Loi Spinetta, and its application décret n° 78-1146 du 7 décembre 1978. The law and its successor texts are codified in the Code de la Construction et de l'Habitation (CCH), articles L.125-1 to L.125-6 and R.125-1 to R.125-21. The professional standard that defines the work of a controlleur technique is the norme NF P 03-100.

Under CCH article R.125-17, contrôle technique is mandatory for certain building categories: ERP (établissements recevant du public) of categories 1 through 4, IGH (immeubles de grande hauteur) where the highest occupied floor is more than 28 metres above ground, and buildings with specific structural or seismic-zone risks. The contrôleur technique is appointed by the maître d'ouvrage and works alongside the maîtrise d'œuvre throughout the design and construction phases, producing an initial report at the end of design and a final report before reception. The major firms operating in this space — SOCOTEC, Bureau Veritas, Apave, Qualiconsult — share a market that has been consolidating around the same statutory framework for nearly five decades.

UK: the Building Safety Act Gateway regime

The UK regime for higher-risk buildings was rewritten by the Building Safety Act 2022, with the Gateway approval process coming into force on 1 October 2023. Higher-Risk Buildings (HRBs) are defined as buildings in England that are at least 18 metres tall or have at least 7 storeys, and contain at least two residential units. Care homes and hospitals meeting the height threshold are also captured.

The Gateway regime introduces three mandatory hold points. Gateway 1 sits at planning permission and requires a fire statement demonstrating that fire safety has been considered in the design. Gateway 2 is the pre-construction approval — a detailed building control approval application made to the Building Safety Regulator (BSR), which must be granted before any construction work can begin. Gateway 3 is the pre-occupation completion certificate. The application at Gateway 2 must include construction control plans, change control plans, mandatory occurrence reporting plans and competence declarations.

US: local plan check under the IBC

The US model is the most decentralised of the three. Plan review (often called plan check or document review) is the responsibility of the local jurisdiction's building department, working from the codes that jurisdiction has adopted. The dominant model codes are published by the International Code Council: the International Building Code (IBC) governs commercial and multi-family construction, the International Residential Code (IRC) governs detached one- and two-family dwellings, and the International Fire Code (IFC), International Mechanical Code (IMC), International Plumbing Code (IPC) and International Energy Conservation Code (IECC) cover the named disciplines. Most jurisdictions adopt these with state and local amendments.

A single commercial submission may be reviewed concurrently against five or more of these codes, plus accessibility under the ADA, plus ASCE 7 for structural loads (adopted by reference into the IBC) and state-specific energy and structural amendments. The review cycle iterates through correction letters — the reviewer identifies non-compliances, the design team revises and resubmits, the reviewer re-reviews — until all disciplines clear. Third-party plan review is widely accepted under IBC Section 104.4 and various state and local amendments, and is heavily used in California, Texas and Florida for high-volume jurisdictions.

Sampling vs full coverage: the model under pressure

The three regimes look different on paper. Operationally, they share a constraint. The volume of regulation each one has to verify has grown faster than the workforce available to verify it. The response has been the same in every market: sampling.

Sampling means reviewing a fraction of any plan in detail and accepting the rest. Specific percentages vary by firm, by project, and by jurisdiction, and no official figure is published — but in practice, only a portion of any given submission is read line by line, and the rest goes to site untouched. This wasn't a deliberate policy decision. It is what scope expansion looks like when it meets margin compression: each project gets less time, each verifier covers more ground, and the inevitable trade-off is depth.

Full coverage is what plan verification looked like before the workforce-to-regulation ratio shifted. Every plan read against every applicable rule. Every dimension checked. Every cross-reference between architectural and MEP drawings followed. Restoring full coverage at modern project speeds is the problem the industry has been chewing on for the past five years, and it is the problem AI-augmented verification was built to solve.

How the three approaches compare

Where AI fits in

AI-augmented verification is what the architecture lets you do once a model can read a plan reliably. The model parses the drawing — identifies walls, doors, dimensions, fire-rated assemblies, room labels, the components of the means of egress — and extracts the data points the verifying experts need to check against the applicable rules. The expert then validates each finding, applies judgement where the regulation requires it, and signs the report.

This is the model Freeda's team works with. Our architects, engineers, accessibility specialists and fire safety experts use proprietary AI to read every page of every submitted document, then validate each finding before issuing the report. The full report goes out in 48 hours, compared to the 100+ hours of manual review the same work requires without AI extraction. Across 2025 our team verified more than 10,000 plans for developers, operators, REITs and inspection firms across France, the UK, the US, and the Middle East. The output is a list of located non-conformities with citations, ranked by impact, with a remediation path for each.

"In construction, every mistake in the plans costs months. Freeda eliminates them from day one." — Peter Starr, Co-founder & CEO of Freeda

The strategic partnership announced with SOCOTEC in January 2026 is the clearest signal that the AI-augmented model is being absorbed into the statutory framework rather than running in parallel to it. From Q1 2026 a Freeda module is being integrated into SOCOTEC's tools and deployed across more than 4,000 pilot projects. The verification industry is not being displaced; it is being expanded.

Where the industry is heading

Three things look settled in 2026. First, full coverage is the bar that every credible verification approach has to meet — regulators across France, the UK and the US have spent the last five years tightening obligations, and sampling at the depth it used to mean is no longer defensible on higher-risk projects. Second, AI is not the differentiator; the experts using it are. The model reads, the expert decides, the firm stands behind the report. Third, the boundary between statutory verification, owner-side QC and external compliance services is becoming less rigid as the SOCOTEC partnership demonstrates. The same underlying technology is being used by bureaux de contrôle, by owner-side teams, and by independent services like ours.

Plan errors caught on paper cost minutes to fix. The same errors caught on site cost months and hundreds of thousands of euros. Compliance verification is one of the few places in a construction project where the economics of catching something early are genuinely asymmetric. Whether that verification happens through a contrôleur technique, through a Gateway 2 submission, through a local plan check, through an in-house QC team, or through a service like Freeda, the point is that it has to happen before construction starts.