The system adapts to you.

Describe the outcome. Lazing builds the working surface around it.

GitHubgit clone https://github.com/MaximilianGerhardt/lazing.git

Local-firstYour runtime. Your data.Private by defaultPersonal Field keeps you in control.Open sourceTransparent. Built for builders.

Lazing mobile command center showing a product launch manifestation

Engines stay scopedCodexClaude CodeOllama

From Intent To Impact

Tell Lazing what should happen. It creates the small system to make it happen.

Lazing begins before the menu, before the dashboard and before the right structure is obvious. You describe the outcome. The system forms around the work.

01 / 08 · Intent

You describe what you want to achieve.

A sentence is enough. Lazing does not ask you to choose the right app first.

02 / 08 · Context

The system reads the field around it.

Personal, organization, client, documents, credentials and boundaries shape what should happen next.

03 / 08 · Method

A proven method enters the moment.

The system can use the logic an expert would normally carry in their head.

04 / 08 · Process

Quality becomes a repeatable process.

The system follows the method, checks the missing inputs and keeps work from becoming vague.

05 / 08 · Engines

Agents do the work they are good at.

Codex, Claude Code, Ollama or local agents can help without owning the whole context.

06 / 08 · Surface

The answer becomes an interface.

A plan, tracker, decision card, workflow, document or workspace appears when the work needs it.

07 / 08 · Resolution

You make the missing decision.

Approve, adjust, delegate or pause before sensitive work continues.

08 / 08 · Trace

The system remembers what happened and why.

Every result can leave a scoped record of context, decision and execution.

Plain Difference

Codex and Claude are engines. Lazing is the cockpit, scope and flight recorder.

The engines can reason, code and compare. Lazing decides which field they may touch, which surface should appear and what must be confirmed before work continues.

Not another chatbot

Chat is one input. The output can be a plan, tracker, decision, workspace or trace.

Engines are not the system

Codex, Claude Code and Ollama do useful work. Lazing keeps scope, surface and memory honest.

A method you can run

Creator expertise becomes a pack people use, not a post people save and forget.

Manifestation Layer

The interface appears when needed.

A chat answer should not always remain text. Sometimes it should become a plan, a decision, a tracker, a review panel, a workspace or a routine that keeps working.

Plan

An idea opens into workstreams, assets, timeline, KPIs and the decisions still missing.

Decision Card

Approve, adjust, delegate or pause without reading another wall of text.

Tracker

A meal, deadline or routine becomes a living state instead of a forgotten note.

Review Panel

Legal, code or strategy work can surface risks, alternatives and consensus.

Workspace

A project gathers context, credentials, documents, trace and next actions.

Fields And Boundaries

Every field has its own memory. Every memory has a boundary.

Personal work, organizations, clients and projects stay separated. Credentials, documents, RAG spaces and agent capabilities belong to the field that owns them.

Personal Field

Private notes, fitness, family, learning, documents and routines stay personal.

Organization Field

Teams, customers, invoices, legal documents and shared workspaces stay scoped.

Scoped Credentials

Access belongs to the field that owns it. Sensitive use asks first.

SQLite Trace

The runtime keeps decisions, changes and agent work understandable.

Pocket Command Center

Decisions should not wait until you are back at a desk.

Send an idea, approve a plan, answer a quick choice, continue an agent, switch a workspace or retrieve a result from your pocket.

Describe your intention

Fix the chat bug and compare Claude Code against Codex.

HeartbeatThe system is alive.

Agent running. Tests pending. One decision needed.

Skills + SOPs

Expertise becomes executable.

Skills hold reusable expertise. Processes hold repeatable quality. Manifestations turn both into interactive work: not just something to read, but something to run.

personal fieldorganization fieldisolated RAGscoped credentialsquick choices

core-workpersonal-fitnessbridge

Start Locally

Run the runtime where the authority belongs.

Lazing is designed for your local machine or VPS. Connect engines when you are ready, keep credentials scoped and let trace stay close to the host.

Terminal

git clone https://github.com/MaximilianGerhardt/lazing.git
cd lazing
./install

View GitHub

For Developers

Many minds. One decision.

Codex, Claude Code, Ollama and local agents are engines. Lazing keeps them scoped, gives them the right surface and records what happened.

View GitHub Read architecture

CodexClaude CodeOllamalocal agentsSQLite tracescoped memoryheartbeatsbenchmark battles

For Creators

Creators should publish methods, not only content.

A creator can turn repeatable expertise into skills, processes, routines and working interfaces that help people get better outcomes.

Build your first pack Apply as creator

A coach turns routines into a nutrition system

An agency turns onboarding into a client workspace

A developer turns review logic into a coding pack

An educator turns a method into a learning journey

Trust, Privacy, Europe

Scope is not a setting. It is product quality.

Local-first operation, no unscoped memory, isolated fields, scoped credentials, consent before sensitive use and traceable events. Designed with European privacy expectations in mind.

FAQ

Plain answers for early builders.

What is Lazing?

Lazing is an adaptive command center. You describe what you want to achieve, and Lazing forms the right working system around it: a plan, decision, tracker, workflow, workspace, document, or traceable resolution.

Is Lazing open source?

Yes. Lazing is planned as an open source runtime and public standard direction, with the website serving as the public home for users, builders, creators, and community.

Does Lazing require Codex or Claude Code?

No. Codex, Claude Code, Ollama and local agents are optional engines. Lazing is the cockpit, scope and trace layer around them.

Can I use Ollama locally?

Yes. Ollama is part of the local-first adapter direction, so builders can run local models while keeping scope and trace inside their own runtime.

Where does my data live?

Lazing is designed for local or VPS operation. The host runtime keeps authority over credentials, scoped memory, adapter capabilities, and trace data.

What is the Manifestation Layer?

The Manifestation Layer is the concept behind turning intent into useful work surfaces. In plain words: an answer can become a plan, decision card, tracker, review panel, workspace or routine instead of remaining only text.

What are Manifestation Packs?

Manifestation Packs combine reusable expertise, required data, risks, confirmations, and quality logic so an intention can become an executable working system.

Is this a chatbot?

No. Chat can be the starting point, but Lazing focuses on forming scoped systems and adaptive interfaces around the work.

Can creators build on it?

Yes. Lazing is designed so creators can turn repeatable expertise into skills, SOPs, routines, lenses, templates, and expert tracks that others can run.

One Interface. Infinite Possibilities.

The surface disappears. The intention remains.

Start locally Star on GitHub Join early builders