WordWield.ai

Independent systems research · Built in Python

Apps & agents.
On common
ground.

Explore the Almasi AI Stack: a filesystem-native runtime and a shared, persistent world of Python objects for applications and AI agents.

Explore the stack Watch the architecture
The architectural ideaShared state
Applications and agents share a persistent object modelAn application and an AI agent connect to the same world of Python objects. That world is built on the Almasi runtime. ApplicationAI agentOne persistent object worldIdentity · Structure · Relationships · StateAlmasi · Filesystem-native runtime
Python objects as a common foundation.
One object model. Fewer boundaries to translate.Runtime structure → Persistent objects → Agent workflows

Three layers, one direction

Meet the Almasi stack.

Each layer builds on the one beneath it, bringing code structure, application state, and agent execution closer together.

01 / RuntimePublic source

Almasi

Your filesystem becomes your runtime.

A Python runtime that resolves library components directly from filesystem structure. Compose and extend libraries through ordinary Python classes, inheritance, and plugins.

Explore code ↗Watch overview ↗
02 / Object worldPublic source

Almasi.o

Persistent Python objects, shared.

A persistent object model unifying identity, relationships, validation, schema, and state. Applications and agents work with the same entities, with fewer representations to synchronize.

Explore code ↗Watch overview ↗
03 / ExecutionIn development

Almasi.do

Agent workflows on persistent state.

The next layer of the stack explores agentic workflows on top of Almasi.o: connecting structured intent, context, and execution within a shared object world.

Discuss the direction ↗

See how it fits together

Architecture, explained.

Two technical walkthroughs, straight from the creator.

Almasi · Runtime architectureFilesystem structure becomes runtime structure
Almasi.o · Persistent object worldA shared foundation for applications and agents

Designed & built by

Alexander Podgorny

AI systems architect. 25+ years in software.
Experience at Google, PayPal, Mozilla, Yahoo, HP, and LG/webOS, spanning hands-on engineering, framework design, and technical leadership.

WordWield.ai is my independent research and development initiative. I designed and implemented the Almasi stack to explore a common foundation for applications and agents — with a natural potential fit for web applications, too.

Meet the architect on LinkedIn ↗

Building in the same direction?

Let’s talk about AI infrastructure, persistent runtimes, architecture, or opportunities to build these systems with a team.

Connect with Alexander ↗