Nemetschek Creator represents Graphisoft’s move beyond conventional cloud-based BIM toward an architecture built specifically for agentic AI. Introduced at Graphisoft IGNITE 2026, the platform combines generative AI with established BIM technologies to support early-stage architectural design and analysis, tells Architosh.

Creator introduces what Graphisoft calls Intent-Based Interactive Design. Architects specify program requirements, site constraints, performance targets, and spatial criteria, and AI agents generate design alternatives that respond to those requirements. The cloud platform can evaluate concepts for daylight, carbon footprint, building-code compliance, and terrain conditions. Designs will eventually synchronize bidirectionally with Archicad.

The underlying software architecture is central to the strategy. Large and small language models do not perform precise geometry calculations themselves. Instead, they act as reasoning layers that call specialized microservices through technologies such as Model Context Protocol. An AI agent evaluating a floor plan, for example, can request deterministic results from dedicated daylight or egress services. Existing C++ B-Rep technology handles precise geometric calculations.

This division addresses a major weakness of generative AI: probabilistic answers can be inappropriate when architects require mathematically reliable results. AI can interpret intent and propose alternatives, while specialized software supplies verifiable engineering and geometric information.

Microservices also allow computational workloads to be distributed. Hundreds of design alternatives could be processed using scalable cloud resources, while smaller models could run locally for private tasks such as building-code checks.

Graphisoft has an advantage because it does not need to rebuild its BIM technology from scratch. Archicad contains decades of C++ geometric mathematics, B-Rep algorithms, and parametric solvers. Its modular architecture allows Graphisoft to refactor these established capabilities into containerized cloud services. Nemetschek Creator therefore combines newer agentic AI with mature BIM computation, creating an architecture in which AI orchestrates design tasks while deterministic software handles calculations requiring precision.

Nemetschek brand Allplan has introduced Allplan Max 2027, a new generation of engineering-focused AI designed to automate repetitive tasks and provide contextual assistance across design, analysis, detailing, and fabrication workflows. Rather than positioning AI as a replacement for engineers, Allplan emphasizes that users remain responsible for design intent, review, approval, and final engineering decisions, says AEC Magazine.

Allplan Max extends AI across the company’s engineering portfolio, including Allplan, Allplan Civil, Frilo, SCIA, and SDS2. The tools target routine activities such as searching for information, checking models, preparing reports, generating formulas, cleaning drawings, and performing repetitive operations.

Within Allplan, Feature Builder allows engineers to describe desired functionality in natural language, with AI and Python used to create customized features. Workflow Builder takes automation further by connecting Allplan with large language models and AI-agent platforms through an MCP Server. Firms can use plain-language instructions to retrieve project information, connect Allplan with other systems, and develop agentic workflows. Other capabilities include AI Formula Writer, AI BIM Analyst, and AI Drafter.

AI also reaches structural engineering. SCIA and Frilo provide AI-assisted analysis and decision support. AI Structural Analyst interprets internal forces, reactions, and deformations to help engineers assess structural results more efficiently. AI Coach provides contextual guidance intended to help junior engineers and occasional software users become productive more quickly.

For steel detailing and fabrication, SDS2 introduces additional AI capabilities. AI BIM Analyst lets users query model information with natural-language prompts, select objects, and generate detailed reports. AI Drafter automatically positions labels on erection drawings while avoiding overlaps.

Allplan Max arrives alongside Allplan 2027, which adds improvements in reinforcement detailing, BIM information workflows, and AI-based customization. Together, the releases show Allplan applying specialized AI across connected engineering tasks rather than limiting it to a standalone assistant.

Snaptrude has developed AECBench, a benchmark designed to measure whether AI models can perform realistic architecture, engineering, and construction tasks. Unlike conventional AI benchmarks focused on reasoning, coding, or question answering, AECBench gives an AI a design brief, asks it to work through an architectural problem, and evaluates the resulting building model.

The benchmark addresses a basic problem facing AEC firms: there is no common way to determine whether an AI system can perform professional design work. Architectural problems rarely have a single correct answer, making traditional pass-or-fail testing inadequate. AECBench therefore divides evaluation into objective core requirements and qualitative design assessments.

Every task checks five fundamentals, including whether spaces overlap, stories stack correctly, all spaces are accessible, numerical requirements in the brief are satisfied, and building elements are parametric components. Additional evaluations cover space planning, BIM detailing, drawings, building services, and value engineering. Deterministic software graders measure objective requirements directly from the model. Design quality is evaluated by an LLM developed from reviews by practicing architects.

Snaptrude also tested its AI agent against Claude Code and Codex. Snaptrude’s agent works directly within a parametric building model, while the coding agents generated Three.js scenes. Snaptrude reported stronger results in areas including adjacency and zoning and BIM detailing, although Claude Code scored higher on site response and value engineering.

The distinction between visual output and usable BIM is central to the comparison. Snaptrude’s agent creates editable rooms, walls, openings, and other building components while maintaining a changelog of design decisions. It can also reason about circulation, privacy, acoustics, and relationships between spaces.

AECBench remains an internal benchmark, so its results should be viewed in that context. Snaptrude plans to open-source the tasks and provide a way for AI labs to run their models through the same evaluation framework.