Kinth AI and the Future of Building Design: One Model From Predesign to Documentation

I have worked in architectural engineering for more than 30 years. I have watched drafting boards give way to CAD, CAD give way to BIM, and energy modeling move from a specialist's afterthought to a code requirement. Few tools have genuinely surprised me along the way. Kinth AI did.

I recently used Kinth on a commercial project, and it did more than I expected. Starting from the project requirements, it developed a coordinated building model covering architectural, structural and MEP systems. I could export that model to Revit and keep working in my normal environment. Inside the same workflow, I could run an energy model, study airflow with computational fluid dynamics (CFD), and check the design against LEED credit requirements.

This article explains what Kinth is, how AI building design of this kind changes the traditional design process, and where architects and engineers still need to apply their own judgment. It is written from a practitioner's perspective, not as a sponsored product review.

The Old Problem: Decisions Made Before the Information Exists

Every building project faces the same uncomfortable truth. The decisions that matter most for cost, energy use and occupant health are made early, during predesign and schematic design, when the team has the least information. Massing, orientation, window-to-wall ratio, floor-to-floor height and the basic HVAC strategy are often set before anyone has run a credible energy model or looked at how air will move through the occupied zone.

By the time detailed analysis is available, usually during design development, changing course is expensive. Simulation becomes a way to document a design rather than to shape it. Coordination problems between architecture, structure and MEP are discovered in clash detection late in documentation, or worse, in the field.

The building performance community has argued for years that simulation should guide design from the start. The current version of the ASHRAE and IBPSA standard on the building performance simulation process, Standard 209-2024, is organized around modeling cycles that begin in early design and continue through design optimization. The difficulty has always been practical: building a separate analysis model at each stage takes time that most fees and schedules do not allow.

What Kinth Is

Kinth describes itself as an AI design automation platform. A user brings in the project context, such as a written brief, reference documents or an existing model, and then describes what should be created or changed. Kinth develops the geometry, coordinates the components and keeps the result editable, with named parts and a record of the design decisions behind them.

The company's public work started in design for manufacturing, and its native CAD export currently highlights SOLIDWORKS, Autodesk Inventor and Autodesk Fusion. Its building-related demonstrations show where the platform is heading. One project study develops a prefabricated, accessible hotel bathroom pod with 14,046 component instances organized into 17 systems, including the steel frame, interior finishes, water and drainage, electrical, and ventilation. Another demonstration builds a coordinated MEP model from a set of public bid documents for an existing building, and others show clash detection and automatic rerouting, material quantity takeoffs, and comparison of the model against as-built field dimensions.

Kinth's own project pages are careful about the status of this work. The bathroom pod is presented as a design study that still needs engineering review before fabrication, and the bid-document model is labeled a coordination model, not a permitted design. That candor is a good sign. It is exactly how a responsible engineering team should present preliminary work.

What It Did on My Project

On my commercial project, four capabilities stood out.

A coordinated multidisciplinary model. Kinth generated architectural, structural and MEP systems as one coordinated model rather than three separate files to be reconciled later. Because the systems were developed together, many of the conflicts that normally surface during clash detection were resolved as the model was built.

Export to Revit. I was able to bring the model into Revit and continue working with the tools and templates my documentation workflow already depends on. For any practice that has invested years in Revit standards, that matters more than any single AI feature. A new tool that forces you to abandon your production environment is a much harder sell.

Energy modeling and CFD in the design loop. I could test energy performance and study airflow while the design was still easy to change. On its public demonstrations, Kinth runs airflow studies with OpenFOAM, a widely used open-source CFD toolkit. It describes its building airflow example as a steady screening simulation rather than a validated comfort or code check. That is the right way to use CFD at this stage: to compare options and catch problems such as dead zones, short-circuiting between supply and return, or drafts near occupied areas, before the ductwork is fixed.

LEED checks. Being able to review the design against LEED credit requirements during design, rather than assembling documentation at the end, is a practical advantage. LEED v5, launched by the U.S. Green Building Council in April 2025, gives half of its points to decarbonization and requires an operational carbon projection and a climate resilience assessment for every project. Earlier energy feedback makes those requirements far easier to plan for. Final certification, of course, still comes through third-party review, not from a design tool.

Why Simulation Inside the Design Model Matters for Healthy Buildings

For SYNKTECT's work in indoor environmental quality, the most important change is not speed. It is that performance questions can be asked at the moment they are cheapest to answer.

Ventilation effectiveness, thermal comfort and daylight are largely decided by early geometry and system choices. When CFD runs on the same model that the architect is shaping, the team can see how a ceiling height, a diffuser layout or an open stair affects air distribution in a classroom or an open office before those choices are locked in. When energy modeling runs alongside it, the team can see the energy cost of each comfort or air quality decision instead of treating energy and indoor environmental quality as competing priorities handled by different consultants at different times.

This is the integrated design process that green building guidance has encouraged for decades. AI design automation makes it practical on projects that could never afford a dedicated simulation team.

From Predesign to LOD 400, and the Road to LOD 500

It is worth being precise about level of development (LOD), because the term is often used loosely. In the BIMForum Level of Development Specification, LOD 400 means a model element is detailed enough for fabrication, assembly and installation, essentially at shop drawing level. LOD 500 is different in kind, not just degree. The 2025 specification states that LOD 500 does not indicate a higher level than LOD 400; it describes a field-verified representation of an existing or as-constructed condition, with its accuracy documented.

That distinction clarifies what AI can and cannot do. A tool like Kinth can carry a model from predesign massing through schematic design and documentation, and in fabrication-oriented work such as prefabricated modules, toward LOD 400 detail. Reaching LOD 500 requires measuring what was actually built. Kinth's demonstration comparing as-built dimensions with the CAD model points toward the next step: keeping one living model that starts as a design, becomes a construction model, and is updated against field measurements into a verified record model for facility management.

If that loop holds up in practice, the payoff for owners is significant. A single model, updated rather than recreated at each stage, reduces duplicated modeling effort, shortens coordination cycles, and gives the building operator a reliable digital record from day one.

A Complete Design Office? Benefits and Limits

After using it, I understand why people describe tools like Kinth as an entire architecture and engineering office in software. For a small firm or a sole practitioner, it provides modeling capacity, multidisciplinary coordination, simulation and documentation support that would normally require a much larger team. That can shift where professional time goes: less time drawing and reconciling, more time thinking about the building and its occupants.

The limits are just as important.

• Professional responsibility does not transfer. Structural, mechanical, electrical and plumbing designs still need review, calculation checks and sealing by licensed professionals. An AI-generated model is a starting point for engineering judgment, not a substitute for it.

• Simulation results depend on inputs. Energy models and CFD studies are only as good as their assumptions about weather, schedules, internal loads and boundary conditions. Screening results should guide decisions, and critical cases should be confirmed with detailed analysis.

• Interoperability needs checking on every project. How well geometry, parameters and families translate into Revit determines how much cleanup the documentation team faces. Test the handoff early.

• Evidence is still young. Most published demonstrations of AI-generated building models, including Kinth's, are design studies. Independent, peer-reviewed evaluation of these tools on completed, occupied buildings has not yet caught up with the technology.

Conclusion

In more than 30 years of practice, I have rarely seen a tool compress the distance between an idea and a coordinated, analyzable building model the way Kinth did on my project. Its real promise is not just automation. It is moving energy, airflow and certification questions to the start of design, when answers are cheapest and occupants benefit most.

The firms that gain the most will treat AI design automation as a powerful extension of professional judgment, not a replacement for it. Use it to explore more options, test performance earlier and coordinate disciplines continuously, then apply the same engineering rigor to its output that you would to any consultant's work.

If you are planning a project and want to bring energy, airflow and indoor environmental quality analysis into the earliest design decisions, SYNKTECT's pre-design studies and design optimization services can help. For related reading, see our articles on Autodesk Forma and healthy building design and free simulation software for healthy and sustainable buildings, or contact us to discuss your project.

Cover image: exploded view of the bathroom pod design study by Kinth. Source: kinth.ai.

References

ASHRAE & IBPSA. (2024). ANSI/ASHRAE/IBPSA Standard 209-2024: Building performance simulation process [Addendum f, 2026]. ASHRAE. https://www.ashrae.org/file%20library/technical%20resources/standards%20and%20guidelines/standards%20addenda/209_2024_f_20260731.pdf

BIMForum. (2025). Level of Development (LOD) Specification, Version 2025, Part I. https://bimforum.org/wp-content/uploads/2024/12/LOD-Spec-2025-Part-I-Official.pdf

Kinth. (2026). Kinth: AI design automation platform. https://kinth.ai/

Kinth. (2026). Bathroom pod: A bathroom design for accessible hotel rooms [Design study]. https://kinth.ai/projects/bathroom-pod/

Kinth. (2026). Demos: Architecture, engineering, construction. https://kinth.ai/demos/

U.S. Green Building Council. (2025, April 28). U.S. Green Building Council launches new, more comprehensive LEED rating system for sustainable buildings [Press release]. Business Wire. https://www.businesswire.com/news/home/20250428258889/en/

SYNKTECT

Innovative design solutions for healthy and sustainable buildings.

Contact

© 2025. All rights reserved. · Privacy Policy

Please email us with your request and contact information. One of our representatives will get back to you shortly.

info@synktect.com