Qentl / Quantum EntanglementArtificial intelligence by Bloch Inc.

Ideas + code + connected intelligence

Think.Build.Connect everything.

Meet Qentl. Bring your ideas, documents, code and AI models into one connected project. See the whole picture in Graphus. Give every task a clear direction.

QENTL.COM / QENTL.AI DESKTOP PREVIEW

01Connected
project map
02Global + focused
prompt scopes
03Private model
profiles

01 / The Graphus workspace

A constellation.
Of your own work.

Each file becomes a point. Folders, functions, ideas and tasks form visible connections. Open a point, inspect the code and follow its development.

◉ GRAPHUS / PROJECT CONSTELLATIONILLUSTRATED PROJECT
Example Graphus relationships between project files, a task and its selected AI modelideas.mdsources.pdfarchitecture.mdsrc/main.pyTask: refineAI modelFILES → RELATIONSHIPS → TASKS → EVOLUTION
01 / KNOWLEDGE

Make ideas usable.

Add documents and images. Turn extracted information and your own ideas into Markdown notes, connected with links and associations.

02 / DEVELOPMENT

See the code evolve.

Open files in the editor, inspect functions, review proposed changes and check supported syntax. Python compilation checks do not execute the project.

03 / ORCHESTRATION

Choose the scope.

Set global project instructions. Focus a task on selected folders, files or functions. Keep its model, context and changed files connected in Graphus.

DESKTOP / PREVIEW

macOS and Windows builds are in validation. The workspace uses your existing Qentl workspace. Project data is stored on the computer running the workspace. Public installer distribution will follow release verification.

02 / Qentl CLI

Your terminal.
The same projects.

Sign in with your Qentl account, talk to the configured models, generate documents and sync a chosen folder with its Graphus project on qentl.ai.

INSTALL / MACOS + LINUX
curl -fsSL https://qentl.ai/cli/install.sh | sh

Needs Python 3.9 or newer. Installs a private environment in ~/.qentl/cli and the command at ~/.local/bin/qentl, without sudo. With pipx on any system, Windows included: pipx install https://qentl.ai/cli/qentl-0.6.1.post1-py3-none-any.whl

SIGN IN / USE
qentl login --email you@example.com
qentl projects list --remote
qentl chat --remote --project PROJECT_ID
qentl sync --project PROJECT_ID --workspace '/absolute/path' --dry-run

Passwords use a hidden prompt and are never saved. Run qentl --help for every command.

CLI / PREVIEW 0.6.1.post1

Unsigned preview package. Read install.sh before running it, or download the wheel and compare it with SHA256SUMS.txt. Accounts are created by invitation; remote commands need an existing Qentl account. Hosted workspaces do not run shell commands.

03 / Models for different tasks

Different models.
One connected project.

Choose a configured model for each task. A model node links to its tasks and their project context, making the work traceable.

PRIVATE PROFILE / 01

Qwen3.5 4B

Compact inference profile on a dedicated GPU.

Structured-response smoke test passed.
PRIVATE PROFILE / 02

Qwen3.5 9B

General project work and programming assistance.

JSON and a basic file-creation agent flow verified.
PRIVATE PROFILE / 03

Qwen3 14B

An additional larger model for comparison across tasks.

Arithmetic, JSON and an exact code-patch smoke test passed.

These profiles have been tested in the operator environment. Availability depends on the configured private service or a local model. Smoke tests are limited checks, not broad capability benchmarks. GPU memory remains local to each machine.

04 / Private by project

Your context.
Your boundaries.

The private Graphus workspace separates projects by user and organization, and keeps project content out of shared training and the public ledger.

  1. 01

    Access by invitation.

    Sign in with your Qentl account. A master invitation creates account access; project owners choose which files collaborators can view or edit.

  2. 02

    A visible scope.

    The model receives the task and its bounded context. Specific task scopes restrict file tools to selected project paths.

  3. 03

    Reviewable changes.

    Compare suggested additions and removals before applying edits. Export project data and control local history retention.

  4. 04

    Clear operating limits.

    Private inference uses configured servers; their hosting region still needs confirmation. The preview is not a claim of global regulatory certification. OS command execution remains disabled in the web workspace.

05 / The Bloch ecosystem

Follow the idea.
Stay close to the source.

Qentl is the new identity of BlochAI. The public chain-data platforms keep their own tools, source coverage and access boundaries.

Distributed compute and contributor roadmap

The existing research fleet has 13 configured GPU workers. This is not an online count or pooled GPU memory. It operates separately from the private model profiles above.

A future “Connect your GPU / Earn BLCH” contributor program remains PLANNED. The program is not active. Eligibility, verification, reward rules and launch timing are pending; no payout is promised.

Quantum Entanglement / by Bloch Inc.

Connect the ideas.
Build what follows.