Skip to content
DAMREV
Start a Structuring Review

Named engine · Application and operations layer

Praxis

The platform's agentic run-time.

Praxis lets software agents do useful work on the platform without stepping outside its controls. Agents use the other engines as tools, but every action routes through Sentinel and Veritas, is signed via Sentry, and composes through the roles held in Stratum. Praxis is aligned with recognised artificial intelligence (AI) governance frameworks, never certified against them.

What it does

What does Praxis do?

Praxis is the agentic AI engine, the run-time of the platform's agentic layer. It runs agents that use the other engines as tools, routed through Sentinel and Veritas and signed via Sentry, composing through Stratum's roles and context. It is aligned with the model of recognised AI governance frameworks, never certified against them.

  • Engines as toolsAgents call the platform's engines rather than working around them.
  • Controls on every actionEach agent step passes compliance and identity checks before it counts.
  • Role-bound agentsAgents act only within the roles and access that Stratum assigns.

How it works

How does Praxis work, step by step?

Three steps, each with one job.

  1. Step one

    Runs agents

    Agents act across the platform, calling engines as tools.

  2. Step two

    Routes through controls

    Each agent action passes Sentinel and Veritas and is signed via Sentry.

  3. Step three

    Composes via Stratum

    Agents work within the roles and operating context Stratum holds.

Where it fits

Which engines does Praxis depend on, and which use it?

Praxis sits in the application and operations layer and draws on four engines. None depends on it; it is the layer that puts the others to work.

Stratum supplies the roles and operating context every agent composes through, so an agent can do only what its role allows. Sentinel supplies the rule check, applying the relevant transfer rules to any action that moves an asset.

Veritas supplies identity, tying each agent action to a verified party. Sentry supplies the signature: where an action needs a user's authorisation, it is signed on that user's device.

Because agents reach every other engine as tools, Praxis inherits their limits rather than adding new powers. An agent that calls Forge meets the same eligibility gate a person would, and an agent that calls Agora meets the same listing rules.

Standards and integrations

Which standards and networks does Praxis work with?

  • ISO/IEC 42001The international standard for AI management systems. Praxis follows its model and holds no certification against it.
  • NIST AI Risk Management FrameworkA framework for managing AI risk from the US National Institute of Standards and Technology (NIST). Praxis is aligned with its model.
  • EU AI ActThe European Union (EU) regulation on artificial intelligence. Praxis is built with its model in view; how its high-risk rules apply to worldwide use remains a reserved question.

Controls

What does Praxis enforce, and what does it never do?

Praxis enforces

Praxis enforces a single path for agent actions: through Sentinel's rules, Veritas identity, and a Sentry signature, within Stratum roles.

Praxis never

  • Lets an agent act outside its assigned role
  • Signs for a user without that user's device
  • Claims certification or compliance under any AI framework

Common questions

What do people ask about Praxis?

Short answers on Praxis: what it does, whether it is AI Act compliant, how agents are kept in check, whether an agent can act for you without your approval, and which engines agents can use.

What does Praxis do?

Praxis is the agentic run-time: it runs agents that use the other engines as tools, routed through compliance and identity.

Is Praxis AI Act compliant?

Praxis is built to the model of recognised AI frameworks, including the EU AI Act, and is aligned with them. It is not certified or compliant.

How are agents kept in check?

Agent actions route through Sentinel and Veritas, are signed via Sentry, and compose through Stratum's roles.

Can an agent act for me without my approval?

Where an action needs a user's authorisation, Sentry obtains it on the user's device. An agent cannot produce that signature alone.

Which engines can agents use?

Agents can call the platform's engines as tools, each under the same rules and gates that apply to a person.

Where next

The pages below pick up where this one ends. Start with the one closest to where your asset or programme stands today.

Last updated:

Ready to build on programmable infrastructure?

Tell us what you want to tokenize. We will show you how DAMREV can run it.