VM4AI · AI engineering

AI work, carried forward.

VM4AI builds AIR, SLUSE and BinLing — different approaches to AI project continuity, governance and execution boundaries. AIR Kit is the open starting point.

Public AIR Kit · private SLUSE development · BinLing research · different licenses

ONE PROJECTTHREE STATES

Focused

Fluid

AIR

→
→
VM4AI portfolio

One umbrella. Different trust boundaries.

These projects address different layers of AI work. They are not a single production-integrated product.

Public · Apache-2.0

AIR Kit

Prompt-based framework for structured tasks, evidence, review and continuity across sessions.

Explore AIR Kit →
Private development

SLUSE

Intercepts and holds protected agent actions. The AIR backend alone decides; SLUSE enforces authenticated outcomes.

See SLUSE status →
Research · version-specific licensing

BinLing

Capsule/payload transport and experimental deterministic runtime research; native SLUSE integration is pending.

See BinLing status →

See the project portfolio and licenses, or inspect work built with AIR.

The problem

Your project shouldn’t reset with the session.

Long AI work creates an ugly choice: stay in an increasingly noisy session, or start fresh and spend time reconstructing work you already did.

Without AIR
Long sessions accumulate noise.
A fresh session means another briefing.
Decisions disappear into session history.
Changing platforms breaks continuity.
With AIR
One active task stays centered.
Structured handoff captures the current state.
A new session continues from that state.
The project can move to another compatible platform.
The AIR system

Focused. Fluid. AIR.

The same project viewed at three levels: the active task, the continuation point, and the project that persists across sessions and compatible platforms.

Focused

One active task at a time.

Focus on one task.

Fluid

Continue, don’t reconstruct.

New Session
Current Session
New Platform

same project state continues

AIR

Designed for continuity across sessions and compatible platforms.

→
→

same AIR project persists

The container can change. The work keeps its state.
Continuity

Work. Handoff. Continue.

AIR carries the explicit working state needed to continue the project instead of asking the next session to reconstruct it from a buried transcript.

01

Work

Keep the active task, constraints, decisions and evidence explicit while the project moves forward.

02

Handoff

Capture the current project state when the session needs to end or the work needs to move.

03

Continue

Load that state in a new session or compatible platform and resume from the project’s recorded position.

Active taskWhat is being worked on now.
DecisionsWhat has already been settled.
ConstraintsWhat the work must respect.
EvidenceWhat supports the current direction.
ApprovalsWhat the user has authorized.
Next stepWhere execution resumes.
Trust & boundaries

Structure helps. Evidence still matters.

AIR is prompt-based and host-model governed. It makes scope, continuity and review boundaries visible, but it does not turn prompt records into independent proof of external events.

Keep the claim proportional

Use AIR’s visible records to inspect the working contract. Use the relevant source, tool, test, operator or backend evidence for claims about the world outside the prompt runtime.

Prompt-based framework, not hidden backend enforcement.
Host capability and compatibility still matter.
External claims still require external evidence.

Pick up the project. Not the briefing.

Use AIR to carry structured project state into the next session or compatible platform.