The Knowledge-based Engineering Workstation
Genworks provides the whole knowledge-based engineering workstation. The engine is Gendl or Genworks GDL: engineering rules as executable code, with parametric geometry, NURBS solids, and web-delivered applications, from the open-source kernel to the commercial Enterprise stack. Readymacs is the console. Basalt is the deployment, declared in one file and brought up with one command on your own machine or your own servers. Lisply-MCP is how AI agents work in the live system beside you.
(define-object robot (base-object)
:computed-slots
((head-radius 1.0)
(body-length 3.0))
:objects
((head :type 'sphere
:radius (the head-radius))
(body :type 'box
:length (the body-length)
:width (twice (the-child length))
:height (* (twice pi) (the head radius)))))What the engine does
Engineering knowledge as executable code: write the rules once, and geometry, analysis, and documentation stay consistent as requirements change.
- Declarative by design
Define objects, not procedures. GDL captures rules and relationships; dependency tracking recomputes exactly what changed, and nothing else. The knowledge stays readable decades later.
- Web-delivered applications
Every model can be a web application: parametric UIs, live 3D in the browser, and MCP endpoints for AI agents, served straight from the running engine with no separate app stack.
- Wireframe to NURBS solids
From open-source parametric wireframe in Gendl to SMLib-backed surfaces, solids, and STEP/IGES exchange in GDL Enterprise. One language across the whole range: commercial capability is an upgrade, not a rewrite.
The whole stack, deployable as code
The workstation is deployable as code: a container stack declared in one file, brought up with one command, with an Emacs console and MCP endpoints so both people and AI agents work in the live system.
Basalt
Free · AGPLThe deployment itself: an Emacs console, Gendl engines on CCL and SBCL, and a monitor, declared in one configuration file. Every service reachable over HTTP, SLIME/Swank, and MCP from the moment it starts.
About Basalt →Readymacs
Free · AGPLThe console: a batteries-included Emacs distribution for AI-assisted Lisp development, whether in a Basalt deployment, standalone in a container, or directly on your machine. Its MCP endpoint lets any agent work beside you in the running editor.
About Readymacs →Lisply-MCP
Free · AGPLThe connective layer: Model Context Protocol middleware that turns any Lisply-compliant backend's REPL into a general-purpose API for AI agents: Claude, Gemini, Codex, or anything else that speaks MCP.
About Lisply-MCP →A deployment's ingress and dashboard roles are filled by commercial binaries, licensed separately.
How the engine thinks
Four ideas do most of the work in a knowledge-based engineering system.
Caching
Once the engine computes something, it does not compute it again to answer the same question.
Dependency tracking
Change an input and everything downstream is invalidated automatically. There is no rebuild step to remember.
Demand-driven
Only the branches needed to answer the question asked are evaluated, out of a tree that may hold thousands of attributes.
Declarative form
You state relationships, not call order. The engine works out the sequence at the moment of demand.
FAQ
Short answers for engineers and buyers.
Where does the design intent live?
In executable code. An engineering rule is written once, as a declarative definition, and the geometry, the analysis, and the documents that depend on it are computed from it. Change an input and everything downstream updates. The rule itself stays readable, and stays yours.
Why Common Lisp?
Its ANSI standard has been stable since 1994, so code written against it keeps running. Its macro system and runtime flexibility suit a knowledge-based engineering system, where the language is extended with new definitions while it runs. Franz Inc has supported Allegro CL continuously since 1984.
What is the difference between Gendl and Genworks GDL?
Gendl is the open-source kernel, under the AGPL. Genworks GDL builds on Gendl with commercial licensing, the SMLib NURBS kernel, HarmonyWare CAD I/O integration, Franz Inc partnership, and enterprise support. Both share the same core architecture, refined since 1997.
How long has Genworks been in business?
Genworks International (founded as Knowledge Based Solutions in 1997) has been operating continuously ever since. We've powered mission-critical applications at Fortune 500 companies since 2002 and maintain support continuity agreements with Franz Inc to ensure long-term stability.
Is this for humans or AI agents?
Both, by design and not by retrofit. Because GDL is homoiconic (code is data) and declarative (you state intent, the engine computes the result), the same define-object source a Lisp developer writes is exactly what an LLM agent emits and rewrites. The runtime is introspectable through the REPL, and the engine is exposed over MCP by Lisply-MCP, part of our free, open-source deployment stack. A human drives it through the editor; an agent drives it through the same tools.