Enterprise digital workforce · Runs locally · Auditable

Digital workers that actually finish the work

Read files, run commands, follow your processes. Tasks are dispatched automatically and results are verified independently. Work runs on your computer by default, sensitive actions need approval, and everything is auditable.

Download Clerk Bit Choose your platform, install, and sign in
Windows client Online installer · 5.0.0.9 Download
Offline package and checksums are in the download center More install options

Windows / macOS · About 1–2 minutes to install

  • Runs locally
  • Approval for sensitive actions
  • Independent verification
  • Enterprise governance
Runs locally; files stay local by default
Approval before sensitive actions
Independent result verification
Tenant isolation and audit
Scenarios

Your work starts here

From a single goal to a verifiable deliverable. Pick a role scenario and see how a digital worker gets it done.

Office work: turn materials into a draft

Upload files, describe the deliverable, and the digital worker organizes the material, writes a draft, and follows your format.

Input Meeting recordings, project files, templates
Deliverables Minutes, weekly report, briefing
Meeting minutes
Weekly report
Briefing
Platform

One platform for the whole workflow

A unified entry point, specialized roles, and an open capability layer — from task creation to delivery.

01

Work entry points

Start tasks from the desktop client, web console, or WeCom.

  • Desktop client
  • Web console
  • WeCom
02

Digital workers

Configure roles, skills, and permissions; multiple agents divide the work.

  • Role configuration
  • Multi-agent collaboration
  • Task orchestration
03

Capability layer

Models, skills, connectors, and file context work together.

  • Skill marketplace
  • MCP connectors
  • Models and runtime
Verification loop

Not a claim of “done” — a deliverable you can check

Every run submits a result, and an independent verification task reviews it. Only a pass completes the task; a fail sends it back.

1

Define

Goal, inputs, and acceptance criteria in one place.

2

Execute

The digital worker reads files, runs commands, and follows the process locally.

3

Submit

Result, attachments, and a self-check are submitted together.

4

Verify independently

Another machine or session reviews it; only a pass counts as done.

Runs on this machine · Fully auditable
Ecosystem

An open ecosystem that keeps growing

Capture skills by role and connect the systems you already use. Capabilities depend on your market and configuration.

Roles

Finance

Reconciliation, reports, expense and invoice checks

Legal

Contract review, clause identification, compliance checks

HR

Resume screening, interview design, onboarding materials

Sales

Account follow-up, material organization, opportunity analysis

Engineering

Troubleshooting, code maintenance, release checks

Operations

Daily reports, competitor monitoring, campaign reviews

Skill examples

Expense report Invoice check AR aging Contract review Clause review Resume screening Competitor research Data query Log inspection Weekly report

Connectors

WeCom Microsoft 365 (SSO) MCP HTTP API SSH / databases Git
Enterprise security

Built for the enterprise, secure by design

From permissions and data to deployment, everything needed to roll out digital workers at scale.

Permission control

Role- and policy-based access with least privilege.

Approval for key actions

Commands, files, and external writes require approval and are logged.

Data boundaries

Tenant isolation; passwords and credentials stay on the machine.

Private deployment

On-premises deployment with your own models.

Knowledge capture

Turn team know-how into reusable Skills.

Controllable cost

Quotas, metering, and billing make cost predictable.

FAQ

Still have questions? Read the docs or contact us.

Where do digital workers run?

By default, on the computer where the client is installed. Files, commands, and session context stay local, and sensitive actions need your approval. You can also run them in a secure cloud sandbox for isolation or centralized management, subject to your configuration.

Which models are supported?

Platform-hosted models and your own NewAPI or OpenAI-compatible models. The exact list depends on your configuration.

Is my data used for training?

No. Data is isolated per tenant and your files are not used for model training.

Can it run on an intranet?

Yes. Private deployment and offline packages are available, and models can point to internal services.

How is pricing calculated?

We offer trial, team, enterprise, and private deployment plans. Contact sales for pricing and quotas.

Put your first digital worker to work today

Download, contact us to start a trial, or book a demo and run through your scenario with us.