Harth
Harth is a browser-based, AI-powered workspace built specifically for architects and architecture teams. It combines frontier large language models and image rendering tools inside a single visual, spatial canvas that is connected to the user's actual 3D building data, rather than treating AI as a separate chat window. The product advertises access to multiple frontier models — including Gemini, ChatGPT, Claude, Grok and Kimi — so teams can switch between models without leaving the canvas. Harth is designed to sit alongside existing CAD and BIM tools rather than replace them. Teams can bring models in from Revit, Archicad, Rhino and IFC, so precision modelling stays in the desktop CAD environment while ideation, research, visualisation and documentation-style tasks move into the web workspace. The platform also supports terrain and existing-building import, 3D satellite terrain for real-site context, live comments anchored over the 3D scene, instant search and actions, and the ability to search the web and convert product images into 3D objects placed in the scene. The platform emphasises agentic workflows: purpose-built agents handle repetitive architectural work, multi-agent workflows sequence several agents at once, and a deep research capability produces substantiated reports on new sites and projects. Because it runs in a web browser, it removes desktop installs, license juggling and version conflicts for the tasks it covers. Pricing details, deployment specifics and technical internals are not published on the homepage. A free trial with no installation and a self-service sign-up are offered, along with a demo-booking option for teams.
- Category
- AI/ML
- Pricing
- Unknown
Problem solved
Architecture teams often juggle many disconnected tools — desktop CAD and BIM software, separate AI chat subscriptions, rendering services and manual site research — which fragments their workflow and slows early-stage design. Harth consolidates frontier AI models, image rendering, 3D site context and agent-driven task automation into one browser-based canvas connected to real building data, so architects can do research, visualisation and repetitive documentation work without switching applications.
Use cases
- Generating and iterating on architectural concept imagery with multiple AI image models
- Running deep research reports on new sites, plots and project briefs
- Automating repetitive architectural tasks with purpose-built and multi-agent workflows
- Importing Revit, Archicad, Rhino or IFC models into a web-based 3D workspace for review
- Placing a project on real 3D satellite terrain with neighbouring buildings for site context
- Reviewing designs collaboratively with comments anchored directly onto the 3D model
- Turning web product images into 3D objects and dropping them into a scene for furnishing studies
- Sketching and ideating on a spatial canvas alongside existing CAD workflows
Audience
- Designers
- Enterprise
- Business Users
- Startups
Pros
- Browser-based with no installation, removing desktop license and version conflicts for supported tasks
- Aggregates multiple frontier AI models (Gemini, ChatGPT, Claude, Grok, Kimi) in one canvas
- Works alongside existing CAD/BIM tools such as Revit, Archicad, Rhino and IFC rather than replacing them
- Combines AI agents, image rendering, 3D site context and research in a single workspace
- Free trial with self-service sign-up and demo booking lowers evaluation friction
Cons
- Pricing and plan details are not published on the public homepage
- No public GitHub repository, so technical architecture and data-handling details are unclear
- Requires an internet connection and browser access, unlike offline desktop CAD tools
- Depth of modelling precision is limited relative to dedicated CAD/BIM software
- Independent benchmarks, reviews and third-party validation are not yet widely available
Official website: https://harth.build/
Canonical LambdaBase page: https://www.lambdabase.com/tools/harth