Demo available

Meshbone

The control plane for AI agents

Run and govern every AI agent and service, across every cloud and model, from one vendor-neutral control plane: what you run, who owns it, what it costs, whether it still meets policy, and which model it calls, so a vendor that changes its terms can be swapped by policy rather than by rebuild.

Subscribe Request demo access Talk to us

The market

AI sprawl is the next shadow IT

As organisations ship more AI agents and services, no one can answer what is running, who owns it, what it costs, or whether it still meets policy. AI sprawl becomes the next shadow IT.

The hyperscalers see it too: Microsoft, Google and AWS are shipping agent control planes. They increasingly reach other platforms, but full control usually runs through their identity, gateway or licence tier, and the record stays in their platform. Most organisations run agents far beyond any single vendor: across clouds, on open-source and self-hosted models.

The result is duplicated spend, ungoverned agents, and risk that scales with every new deployment, exactly when control matters most. What is missing is one neutral, sovereign place to govern them all, and one place to change the model behind an agent when a provider changes its price, its terms or its region, without touching the agent.

How Meshbone helps your organisation

One vendor-neutral control plane

Meshbone is designed to run and govern AI agents and services from one place, across clouds and models, with no lock-in to any one stack. Today its gateway supports agents built on the Anthropic, OpenAI and Gemini SDKs. An agent calls the gateway, not the vendor, so the model behind it is a policy setting: approve the model creators you trust, pin a version, and route to another when a vendor changes its terms.

Know what you run, who owns it, what it costs

A complete, live inventory with ownership, cost and policy-conformance in one view across your entire AI estate.

Govern agents beyond any one platform

Meshbone is designed to govern agents from any platform in one place, with the record kept by you.

Sovereign and policy-enforced by default

Run in your own country and your own cloud, with governance, guardrails and runtime blocking enforced continuously from first request to retirement.

What Meshbone does

Platform, data, and AI engineering leaders running AI across clouds and vendors at scale.

Vendor-neutral, multi-cloud control plane (Microsoft, AWS, Google, open-source, self-hosted)
Unified registry of AI agents and services with ownership and cost
Model route as policy: approved creators, pinned versions, provider, model, region and cost recorded on every call, swap without a rebuild
Continuous policy enforcement, guardrails and runtime blocking
Cost and policy conformance in a single view
Sovereign / in-region deployment
Lifecycle management from request to retirement
Shadow-AI discovery and monitoring, with drift monitoring planned

How every deployment is structured

Standard rails. Made-to-measure services.

The governance rails are ready-made and the same for every customer, so their assurance is reused: assessed once, carried into every service, while each service is still assessed on its own. The services that run on them are built to fit the way your organisation actually works. You get software that fits, and it is still governed.

Before the build. Before anything is built, Orbit Roadmaps maps where AI adds value across the organisation and sequences the work into a costed, phased roadmap. The build follows the roadmap, not the other way round. Orbit Roadmaps

1This product

Governance rails

Ready-made. Identical for every customer.

AIG SentinelMeshboneagentVIKING™

AIG Sentinel sets the policy and holds the evidence. Meshbone enforces it at run time on every agent and service. agentVIKING™ is the set of rogue-agent controls the rails enforce on every agent they run. Because the rails are the same everywhere, an assessor who has assessed them once need not assess them again for the next service. Each service still carries its own assessment of its purpose, data, risks, integrations, rules and model use. The rails make that assessment smaller. They do not replace it. The rails subscription is the derisking spend: the controls, the monitoring and the evidence that the AI tools themselves do not include, paid for once and reused by every service rather than rebuilt around each one.

Rails subscription, on every deployment

Service templates

Productised. Drawn from the catalogue.

Nanolitte

Nanolitte is the build engine: every template is a BPMN 2.0 process it compiles into governed agents, with the guardrails inside. Grants, cases, permits, procurement, records, assets: each starts from a template that already runs on the rails, with intake, assessment, decision and record built in. A new service begins most of the way there rather than from a blank page.

Subscription per service

Made to measure

Built for you. Kept small by design, measured on every build.

NanolitteDataballastExecdive

Your rules, your data model, your thresholds and your approval chain, configured and coded onto the template. This is where the hard part of any regulated service lives: the legislation, the exceptions, the delegations and the systems you keep. So we scope that share in three columns before a build starts, standard, configured and coded, and report it in the same three when it ends. Anything a second customer could use goes back into the template.

Fixed-scope build

4This product

Run and support

Ongoing, on the rails.

Meshbone

Hosting, monitoring, policy re-attestation and upgrades, in your region. After go-live your own team changes a threshold or a form within guardrails, without waiting on us.

Subscription

The rule

No made-to-measure build without the rails underneath it. That is the rule that keeps a custom service governed, upgradeable and provable, rather than a fork somebody has to maintain by hand forever.

agentVIKING^™

Guarded by agentVIKING™

Any agent routed through the gateway, ours or yours, under the same four classes of control

Directive, preventive, detective and corrective. Written in AIG Sentinel, enforced at the gateway in front of the model, carried in every Nanolitte template, evidenced on one chain.

Meshbone: enforces it at run time. The enforcement point. Its gateway sits in front of the model, so the policy AIG Sentinel issues, content inspection, run-time blocking, the kill switch and quarantine apply to any agent that calls a model through it, built on Arrochar Labs or not.

See the controls

Directive

What the agent is allowed to do, as data.

Preventive

What stops it before it acts.

Detective

How we know when it goes wrong.

Corrective

How it is stopped, contained and fixed.

How it fits together

A complete suite of AI governance agents. An architecture that makes them accountable.

Each part of ASG™ does one job, in order, and hands the next part something it can check. That is the difference between building AI agents and running them so an auditor, a regulator or a minister can see what happened.

Before the first build, Orbit Roadmaps finds where AI creates value and profiles the risk.

After it, Execdive turns what the chain recorded into plain-language insight for executives.

Policy, enforcement, templates, data, controls, evidence. Then services.

This describes the architecture, not what every deployment runs today: 7 of the 30 agentVIKING™ controls are still in development.

One architecture, every product

A full AI-agentic experience

Every Arrochar Labs product is built the same way: agentic, event-driven, and governed with human-in-the-loop controls, so AI does the work while your people stay in control.

Agentic

AI agents do the work, end to end. Every product runs on autonomous agents that take action across your processes, not just a chat box, but software that actually gets the job done.

Event-driven

It reacts in real time. Products respond to what's actually happening (new data, a policy breach, an incoming request, a change in your estate) instead of waiting on a batch run or a person to press go.

Human-in-the-loop

Your people stay in control. Approval gates, oversight and a tamper-evident audit trail keep every agent accountable, so automation never runs unchecked.