Get started
01 / 06WorkspaceFive-mode preview · read-only

Vitna Workspace

The operating environmentfor your company’s AI.

A workspace for questions, code and reusable workflows.
A control plane for the models and policies behind them.

Explore the Vitna Workspace concept

Vitna / WorkspaceWorkspace concept preview

Build / Agents & apps

Your task

Build me a competitor-monitoring agent for Anthropic, OpenAI and Google.

An agent you can inspect.

Turn an intent into a workflow with its sources, tools and authority in view.

Competitor monitorDraft for review
Policy for this taskresearch-reviewAuthority stays with the task
WorkflowDraft Review
Agents & apps

Competitor monitor

Proposed workflow · Draft for review

Planned sources
AnthropicOpenAIGoogle
  1. Watch the official sources

    Read product announcements and published model changes.

  2. Compare what moved

    Compare the new reading with the previous brief.

  3. Prepare a sourced update

    Draft to Competitive Intelligence. Hold publication for review.

    Human review before publication
Illustrative draftNo task executed
Model lane vitna/researchAuthority Scoped to the task
Execution details

The proposed infrastructure view for this task.

Model lane
vitna/researchModel resolved by an accepted policy. Fallbacks must be declared.
Policy
research-reviewIllustrative policy name
Tools & context
Public web · Approved company contextProposed connections
Permissions
Read approved sources. Draft to Competitive Intelligence. No external actions.
Cost
Not measuredNo run has executed. Actual usage is recorded after execution.
Evaluation
Test candidate models on this workflow before changing its policy.
See the existing request ledger
Illustrative workspace. No tools connected.See the product architecture
02 / 06One environmentUse · Build · Control

One environment

The work in front of you.
Every layer within reach.

The workspace concept brings everyday tasks and builder tools together, with the context and authority of each task kept in view.

Use AI

A workspace for the task.

Ask questions, work with files and follow an answer to its sources.

Workspace concept
Build with AI

From intent to working artifact.

Carry the task into a coding session or a reusable workflow. Agents and generated apps are proposed extensions.

Code: pre-release
Control how AI runs

Choose with evidence.

Compare models, review a decision and deliver its signed policy to your runtime.

Open the control plane
Available today
03 / 06Decision loopAvailable today

The infrastructure in view

Do the work.
Know what runs beneath it.

A model choice should carry its reasoning. The decision engine compares market, evaluation and production evidence against your workload’s constraints.

Stay, test or switch. Review the decision before its policy reaches the work. Connecting this loop to the workspace is the next step.

See how a decision is made
The decision loop
Market recordEvaluation evidenceProduction readingsYour constraints
Decision engineA reason to act.
Or a reason to stay.
StayTestSwitch
The existing decision loop is the foundation for the proposed workspace.
04 / 06Product architecture9 layers · concept to runtime

The product architecture

The work above.
The control plane underneath.

A connected environment, built on infrastructure you can already use.

01 The experience

Workspace · Code · Agents · Apps

One task, from an initial question to a reviewed artifact.

Workspace concept
Code: pre-release
Agents and apps proposed
02 The context

Knowledge · Connect

Company context and tool access, scoped to the work.

Workspace integration proposed
03 The foundation

Models · Decisions · Runtime

Sourced evidence, accepted policies and a request path you own.

Available today
Explore all product layers and availability
Vitna / The layers of the environmentCurrent product status

Scroll the table to see every layer’s status.

LayerIts jobToday
WorkspaceChat, files, artifacts and task sessions in one surface.Concept preview
CodeEngineering sessions with action approvals and run receipts.Pre-release
Agents · AppsPersistent tasks, reusable workflows and internal tools.Proposed
Knowledge · ConnectCompany context and scoped access to tools and systems.Workspace integration proposed
ModelsA public catalog with sourced prices and published evaluations.Explore the catalog
Router · PoliciesNamed workload lanes, accepted choices and declared fallbacks.Runtime available
ApertureEvaluate whether a workload should stay, test or switch.Try the Decision Lab
Observe · ControlDecision history, policy provenance and request usage.Register and local ledger
DeployDeliver accepted policies; later, promote agents and apps.Policy delivery available; agent deployment proposed
05 / 06Delivery pathsConfig · Gateways · Signed policy

Use the infrastructure today

A decision.
Delivered your way.

Keep your gateway. Load a config. Or run a signed policy on infrastructure you own.

Use the record on its own. Add the decision loop when you need it.

06 / 06Evidence and closeSource · Approval · Runtime

Evidence by design

Every change has a trail.
Every trail can be checked.

Start with the figure. Follow it to its source. Then follow an accepted decision to the policy that applied it.

Trace the source

Follow a published price to the source document and the signed observation.

Verify a figure

Keep approval explicit

Review a proposal. Accept it as a separate step. Deliver the accepted policy.

Review decisions

Own the request path

Your runtime reads a signed policy from disk and records requests locally. It makes no call to vitna.ai while serving.

Read the runtime contract

Start with one workload

Your work. Your models.
Your control.

Explore the workspace direction. Use the decision tools today.