Skip to content

KVARK Platform

One foundation to connect, control and run AI.

AI Services, AI Hub and AI Development share KVARK Platform. It connects your systems, applies your rules, logs what users and agents do, and manages models and costs.

It connects your systems, applies your rules, logs what users and agents do, and manages models and costs.

Business value

Add AI use cases without starting from scratch

You set up integration, governance, cost controls and operations once, and every supported application shares them. Each new AI use case builds on them, so it takes less time to deploy and costs less to add.

Integration and ingestion

Connect your source systems, such as Teams, Outlook, SharePoint and SQL databases. KVARK takes in their data and files and indexes them for search.

Connect source systems such as Teams, Outlook, SharePoint and SQL databases. KVARK indexes their data and files for search.

  • Recognise and enrich what the files contain.
  • Manage metadata and set up search.

Example · source systems

Data classification and retention

Handle data by how confidential it is, and identify personal data (PII) in it. Set how long KVARK keeps information, to support your regulatory obligations.

Handle data by confidentiality and identify personal data (PII). Set retention rules, to support your regulatory obligations.

Sketch · a confidential memo

Permission management

Access follows the policies and permissions set in your source systems, your identity and access management (IAM) or both. That helps close gaps in security and governance.

Access follows permissions in your source systems, your IAM or both. That helps close security and governance gaps.

  • Grant access by role (RBAC) or by attribute (ABAC).
  • Manage users, roles and permissions in KVARK too.

Example · roles by source

Tracing and logging

Log what users and agents do, the tools they call and the models’ states, so you can check what happened. Trace outputs back to their original sources.

Log what users and agents do, their tool calls and model states. Trace outputs to their original sources.

  • Set up logging and read audit reports.

Example · a chat log

Guardrails, approvals and sandboxes

Set AI policies and instructions for your organisation, a role or a scenario. Agents can pause for an employee’s review or approval, so you decide what they may do.

Set AI policies by organisation, role or scenario. Agents can pause for employee review or approval, and run in isolated sandboxes.

  • Restrict requests and actions that clash with your policies or regulatory requirements.
  • Run agents in isolated Agent Runtime Sandboxes.
  • Detect and mask personal data.

Sketch · paused for approval

Business context and memory

Manage the business context of your organisation and its people, by role and operating model. Memory under your control helps you keep what your teams know.

Manage context by role. Memory you control helps you keep what teams know. Gluomem, the cross-platform memory, is part of KVARK Family.

  • Control each memory’s owner, source, access scope, retention and deletion.
  • Gluomem, the cross-platform memory, is part of KVARK Family.

Sketch · context and memory

Platform and cost management

Track calls to third-party services, applications and language models, and see what they cost in total. Set usage limits, so you control what you spend.

Track calls to third-party services, applications and language models, see their total cost and set usage limits.

  • Manage licences, updates, backups and archiving.
  • See usage in reports and analytics.

Sketch · costs and limits

Models and AIOps

Manage your local and approved external models through their lifecycle, including KVARK Sparkle and Flash from KVARK Family. Route each task to a suitable model, which helps keep costs down.

Manage local and approved external models, including KVARK Sparkle and Flash from KVARK Family. Model routing helps keep costs down.

  • Decide which tools agents can use.

Example · models by role

Business value

Add AI use cases without starting from scratch

Set up integration, governance, cost controls and operations once. Supported applications share them, so new use cases deploy faster and cost less to add.

Start with one use case. Run it under your rules.

We show KVARK on your kind of documents, your access rules and your deployment constraints. Most teams start with a four to six week analysis and a six week pilot.

We show KVARK on your kind of documents, access rules and deployment constraints.