Lytos for organizations

Lytos is not another SaaS to integrate. It’s an open-source method that puts your projects’ context inside your Git repos — where it always should have been. This page lists what that changes concretely for an organization, without technical jargon.

1. Measurable productivity

Today, your developers re-explain their project at every AI session — the architecture, the conventions, the constraints. This friction repeats dozens of times a day across a team.

With Lytos, the project’s context (manifest, rules, memory) is read automatically by the AI at the start of every session. No re-explaining. No generic code to fix afterwards.

The result: less time wasted re-framing the AI, more time on what matters — design, validation, architecture.

2. Sovereignty over your project context

With classic AI tools, your project context — your conventions, your technical decisions, your business rules — ends up on the vendor’s servers, in their format, dependent on their API.

With Lytos, everything is Markdown in your Git repo: your manifest, your rules, your project memory. Versioned, portable, readable by any model.

When a vendor changes prices, switches models, or disappears, your context doesn’t move. It belongs to you.

3. Auditability: AI stops being a black box

A private AI chat leaves nothing behind. No trace of why a piece of code was generated, under which rules, from which spec.

With Lytos, every line generated by AI rests on:

a versioned issue (the what),
explicit rules (the constraints),
a referenced skill (the how).

Everything is in the repo, readable cold. Audit reviews, compliance checks, post-incident debriefs — all become traceable. Particularly relevant in GDPR, ISO 27001, or regulated-industry contexts.

4. Business continuity: switching AI without losing everything

AI models evolve every 3 to 6 months. Vendors change pricing, APIs, terms. Claude today. GPT, Mistral, or a newcomer tomorrow. Without preparation, every migration is a project of its own.

Lytos is model-agnostic. The manifest, rules, and memory work identically with Claude Code, Cursor, Codex, Copilot, Gemini, Windsurf — and whatever ships next.

Switching AI vendor becomes an operational choice, not a strategic risk.

5. Accelerated onboarding and junior ramp-up

Today, onboarding a new developer takes weeks: implicit context transfers, code reading, endless questions to seniors. With AI as a complement, the junior or new hire produces code… but not in the project’s style.

With Lytos:

The new dev reads the manifest and knows the project in minutes.
The AI produces within the project’s frame from the very first issue (rules, skills, and memory act as silent senior guidance).
Senior time is preserved: fewer repetitive questions, more time on what matters.

Concretely, a junior with Lytos reaches operational level faster, and their output looks like a senior’s — because the frame is shared.

How Lytos fits without disruption

Lytos doesn’t replace your existing tools. It adds to them:

Your product ticketing (Jira, Linear, Azure DevOps) keeps its role for product needs.
Your CI/CD, your code reviews, your conventions stay in place.
Your developers remain developers — Lytos simply adds a framing skill to their craft: prompt engineering applied to the project.

The PO describes the product need in Jira. The developer translates it into a Lytos issue (technical spec for the AI, in the project’s vocabulary). The AI implements within the frame. The lead dev reviews. The pipeline deploys. Nothing changes in the chain — a quality link is added.

See the detailed team workflow →

Getting started

Lytos is open-source, MIT-licensed, telemetry-free. No commitment, no account to create, no SaaS to approve.

npm install -g lytos-cli
lyt init

Three minutes to scaffold the structure inside an existing project. A developer can test it locally. A team can adopt project by project, without disrupting the organization.

To train a new arrival quickly — a human or an AI agent — the lytos-learn repo provides a pre-configured project (a todo API) with real issues to solve. In 7 steps, the method is learned hands-on, without tying up senior time.

In summary

Stake	Without Lytos	With Lytos
Productivity	Re-explain at every session	Context read automatically
Sovereignty	Context at the vendor	Context in your Git repo
Auditability	Private, opaque chats	Issue + rules + skill traceable
Continuity	Vendor lock-in	Model independence
Onboarding	Weeks of implicit transfer	Manifest read in minutes

Lytos is a pragmatic method, not a revolution. It formalizes and makes governable what many teams already do implicitly. The entry cost is low; the impact measures quickly.

To go further:

The method in detail — the 5 pillars
The Lytos team workflow — how it fits in, step by step
Compatibility with your AI tools — native adapters for Claude Code, Cursor, Copilot, Gemini, Windsurf, Codex