Practal

Practal — Designing (with) Precise Abstractions

A foundation for mathematics that works the way you think

Abstraction logic unifies the single mathematical universe of first-order logic with the general variable binding of higher-order logics — yielding a complete logic with elementary semantics. Practal is the proof assistant that brings it to life.

Try Practal Zero Learn More

Available now

Practal Zero runs entirely in your browser

Practal Zero is the first public Practal preview. No installation, no backend, no accounts. Declare or define abstractions, state axioms, and work with checked theories — all in abstraction logic.

This first version is deliberately local and simple. AI integration, live file sync, and collaborative cloud features are planned for later.

Zero starts with a small introduction document so the workspace opens quickly. Larger example theories are available as a separate download for import/export experiments.

Open Practal Zero Download Examples

Planned AI-Assisted Proving

Designed for future AI collaboration

The current Practal Zero preview does not include AI integration or live file sync. The longer-term design is still file-centered: an AI agent should work on the same theory files as you, and every generated proof step should be checked in the same way as your own work.

1
You write in Practal Zero's structured editor — declaring abstractions, stating axioms, sketching the outline of a theory.
2
The AI reads the same theory files through local tooling. It sees your declarations, your goals, and eventually dedicated interaction blocks that provide context and specify what needs to be proved.
3
The AI writes proof steps and supporting edits back into the files. Practal checks the result through the ordinary kernel.
4
You continue — refining, redirecting, or going deeper. The AI becomes a collaborator on your files, not a black box.

This direction is promising because abstraction logic is designed to be explicit and regular. The goal is to make the barrier to entry for an AI as low as it is for a human.

For now, use Practal Zero as a standalone browser workspace with local browser storage and manual import/export.


The Idea

Why another foundation?

Traditional logics force a choice. First-order logic gives you a single universe of mathematical objects but no way to bind variables beyond simple quantification. Higher-order logics give you general variable binding but split the universe into an infinite tower of types. Abstraction logic refuses the trade-off.

Logic Single Universe General Binders
First-Order Yes No
Second-Order No No
Higher-Order No Yes
Abstraction Yes Yes

The key insight is that just like in first-order logic, but unlike in higher-order logic, operations need not be mathematical objects. An operation takes values from the universe and returns a value. Crucially, an operator takes operations and returns a value. This avoids inconsistency by sidestepping Cantor's theorem without sacrificing expressiveness. The resulting semantics is elementary and based on abstraction algebras with a logical order and a valuation space.

Expressive

Classical, intuitionistic, modal, paraconsistent and quantum logics; lambda calculus; dependent type theory; algebraic effects — these are all expressible as abstraction logics.

Complete

Every abstraction logic based only on pure deduction is complete with respect to general models, and even with respect to classical models.

Simple

The syntax and semantics of abstraction logic are simple and elementary. This gives working with it a similar feel as computer algebra.

Open

The logic is open and welcomes scrutiny. Practal Zero is free to use, and deeper local tooling is being prepared.


What can it express?

One logic, many theories

In abstraction logic, you don't switch systems to work with a different logic. Abstraction logic can adapt to many different logical paradigms and fits many mathematical situations. You declare or define abstractions, state axioms as needed, and reason within the resulting theory. There is no distinction between logical and mathematical constants, everything is just an abstraction.

Paperweight example library

The larger example library is available as a ZIP instead of being loaded into every fresh Practal Zero workspace. Import it when you want to explore substantial theories or try Zero's import/export workflow.

Download Example Theories
The archive contains current Practal files plus a README. It checks as a standalone library with no errors and no warnings.

The Software

One logic, staged tools

Practal is built in layers. The public release available now is Practal Zero, a free browser preview. Practal Code, editor integrations, AI-assisted workflows, and Practal Cloud are still under development.

Internal for now
Practal Code
Local tooling layer

The local kernel and tooling layer behind Practal. It is being used to build the current system, but it is not a public product or open-source release yet.

  • Kernel and document processing
  • Command-line and editor-integration foundation
  • Used internally by Practal Zero and experiments
  • Public release shape still to be decided
For tool builders, AI researchers, and power users
Coming later
Practal Cloud
Planned hosted service

The future cloud platform for teams and institutions. Shared theory libraries, persistent workspaces, collaboration, and managed AI support belong here later.

  • Collaborative editing and shared workspaces
  • Public registry of importable theories
  • Cloud-hosted AI-assisted proving
  • Team management and access control
For teams, institutions, and professional users

Updates

What’s happening

Latest work and blog posts. Check back soon, or subscribe via RSS.

April 28, 2026 Code Paperweight
January 25, 2025 Extended Abstract Abstraction Logic Is All You Need
January 22, 2024 Text Format Recursive teXt

Support This Work

Let’s build the foundation

Practal is an independently funded project. The logic is open, and Practal Zero is free to use. If you find the ideas compelling, you can support continued development through GitHub Sponsors. Every contribution helps move the work forward.

Sponsor on GitHub Get the Book