AI that fits the way your organization operates

Put AI to work.
Keep the control.

ARF gives employees and software teams one place to work with AI while your identity, data, infrastructure and operating rules stay connected.

Talk to our team
01
Your data stays in your boundary
02
One policy layer for every team
03
Deploy on infrastructure you choose

DATA SOVEREIGNTY

Work

Code

Control

ARF / COREACTIVE

Enterprise Intelligence,
under your control.

SECURE CLOUDFULL AIR-GAP

✥ Move to inspect

Built for day-to-day adoption

Everything teams need to
use AI responsibly.

Scroll to explore

01 · ARF-WORK

A shared AI workspace for everyday work.

Help teams ask, write, analyze and act on approved knowledge without moving work into unmanaged tools.

02 · ARF-CODE

A coding agent that understands the repository.

ARF Code works from repository context to help engineers plan changes, write code, run through reviews and prepare tests.

03 · Data sovereignty

Business context stays protected.

Keep documents, source code, customer information and internal knowledge within the environment your organization controls.

04 · Centralized governance

Govern access, spend and change from one place.

Use SSO and SCIM, set model and tool policies, manage limits and retain an audit record that cannot be changed.

05 · Deployment choice

Choose the operating model that fits your risk profile.

Run ARF in your own environment or use an isolated Secure Cloud deployment with a dedicated boundary.

AI without the blind spots

Give teams AI access.
Keep every decision visible.

ARF connects the AI workspace to the controls your organization already needs: identity, permissions, usage limits, approvals and audit.

AI without organizational control

Useful answers. Unclear accountability.

  • 01 Where do prompts and files go?
  • 02 Who can access each model and tool?
  • 03 What record remains after work is done?
ARF / Control

ARF inside your environment

One AI platform, managed as part of your organization.

ARF Workspace, Control, Norm and Core connect everyday AI work to the policies, models and infrastructure behind it.

WORKCODEPRIVATE INFRASTRUCTURE

One platform for business and engineering

Useful for every team.
Deep enough for engineering.

ARF Work gives the wider organization a practical AI workspace. ARF Code gives engineering teams an agent that works with real repository context.

01

ARF-WORK

For business teams

Turn everyday requests into finished work.

Use approved business knowledge to answer questions, analyze documents and move routine work forward without losing context.

  • → Ask and investigate
  • → Write and analyze
  • → Use approved knowledge
02

ARF-CODE

For software teams

From repository context to pull request.

ARF Code helps teams understand a codebase, plan changes, write and review code, and prepare tests and documentation in the same governed environment.

  • → Understand the repository
  • → Plan, write and review changes
  • → Prepare tests and documentation

ONE SECURE PLATFORM

Work→Code→Control

One control layer, from the workspace to the infrastructure.

Controls that support adoption

AI use you can
see, manage and audit.

Connect ARF to your corporate directory, define access and approval policies, set usage limits and keep an immutable record of activity.

01

Start with your existing identity system.

Use SSO, SCIM and organization-level roles to give the right people access without creating a separate identity layer.

02

Set the rules for models, tools and agents.

Apply policies by user, team or organization and require approval where work needs a human decision.

03

Understand use, cost and activity.

Track adoption and resources with analytics, limits and an immutable audit trail.

Deployment that matches your environment

Two ways to deploy.
One governed platform.

Choose the operating model that fits your infrastructure, security requirements and procurement path. ARF keeps the product experience and control layer consistent.

01FULL AIR-GAP

ARF SOVEREIGN

Run ARF where your infrastructure already lives.

Deploy on your own NVIDIA GPU servers for on-premise and Full Air-Gap operation. Model execution, data and policy stay inside your environment.

ON-PREMISEFULL AIR-GAP
02SECURE CLOUD

ARF SECURE CLOUD

Get started without building GPU infrastructure first.

Use a dedicated, isolated environment operated by Konvo while keeping the policy, audit and product experience you expect from ARF.

DEDICATEDISOLATED

Model and compute

ARF Matrix 1 for agentic work.
Capacity planned around your organization.

ARF Matrix 1 brings multimodal, agentic capability to controlled enterprise environments. Capacity references support planning; final sizing follows your workload and deployment needs.

ARF Matrix 1 · Multimodal agentic model

762B parameters, optimized for enterprise work.

Designed for controlled environments, ARF Matrix 1 supports multimodal and agentic workloads inside the boundary your organization defines.

Total parameters
762B
Deployment mode

Access channels

Web · Desktop · Mobile · CLI / TUI · VS Code extension

100+ ready integrations

Capacity reference

Choose your compute.
See its capacity.

These reference configurations are planning inputs, not capacity limits. Final sizing is assessed with your workload and deployment requirements.

Compute profile
8× A100 (80GB)

Published reference for 8× A100 (80GB): 2400 recommended total users and 1600 active sessions.

Recommended total users
2,400
Active sessions
1600

Reference values only — capacity changes with workload and deployment design.

Independent benchmarks

Strong where
agentic work is measured.

ARF Matrix 1 ranks in the top three across seven of eight published evaluations.

Terminal-Bench 2.1Rank 3
ARF score
87.9
Highest score
88.8
FrontierSWERank 2
ARF score
77.9
Highest score
88.2
PostTrainBenchRank 2
ARF score
39.7
Highest score
41.8
SWE-Marathon v1.1Rank 3
ARF score
42.4
Highest score
48.8

Top three in seven of eight evaluations.

Questions, answered

The important details.
Made clear.

What is ARF?

ARF is an enterprise AI platform that connects employee and engineering workflows to governed access, policies, models, inference and infrastructure.

What is the difference between ARF-WORK and ARF-CODE?

ARF Work supports everyday business work in a shared AI workspace. ARF Code gives software teams an agentic environment that works with repository context, code, tests and documentation.

How is ARF deployed?

ARF is available as ARF Sovereign on your own infrastructure, including Full Air-Gap, or as ARF Secure Cloud in a dedicated isolated environment.