PromenautThe platform
The operating model built for AI, in one platform.
Promenaut maps, architects, coordinates and controls your hybrid workforce, with the oversight and control a regulated institution needs built in. This is what it is built from: nine layers, from the work people see to the estate it runs in.
Scale AI out of the silo.
Out of the pilot
Take a workflow that works in one team and run it across the organisation: operations, compliance, finance and change, not just engineering.
One operating layer
Workflows run on the same platform, under the same controls, and are found in one registry. No new silo per team, no per-tool overhead.
Gains that compound
What one workflow records in the knowledge graph, the next can draw on. Value accrues to the organisation instead of splintering across disconnected tools.
Nine layers. One governed workforce.
The same map Promenaut’s documents use: what people use at the top, what runs and governs the work in the middle, and what it is built on underneath. Choose a layer.
The workforceWhat people and systems use.
The platformWhat runs and governs it.
The foundationWhat it is built on and built with.
Layer 7 · Control
Identity, tenancy, permissions, approvals and decisions, model risk, audit, cost and budgets.
What it is not. Not a layer added afterwards. Approvals, audit, model risk and cost are recorded by the work as it runs, not reconstructed from logs later.
Approval gate on agent work
Work can be held for a human decision, approved or rejected individually or in bulk, with the decision history kept and retrievable per workflow.
A diagram of one run for auditors
An auditor can open a whole workflow run from the run's page in the Workflow Lab, read-only, with the run's verdict laid over the workflow's teams and agents: an agent whose run failed, or that the roster check did not cover, is marked on its card with a text label, and unexpected agents and calls to service teams that were not declared are listed above the canvas.
What this will cost, before you run it
Before a workflow is launched the platform resolves the agents it would run, sums their credit tiers, converts the total to money at the enterprise contracted rate, and reports whether it fits the annual pool and what proportion of the pool it would consume.
Start with the question you are asking.
Security, oversight, cost, reliability, engineering or adoption: each has its own answer, drawn from the layers behind it, and its own document.
Oversight and model risk
Can we govern it, and show that we did?
Answered by Control and Agents.
The full answer for model risk and second line →Control
Approval gate on agent work
Work can be held for a human decision, approved or rejected individually or in bulk, with the decision history kept and retrievable per workflow.
Agents
Per-agent control plane
Each agent has one control surface covering identity and entitlement, contract and intent, model and behaviour, capabilities, governance, lifecycle and versions, control and containment, and assurance and cost.
Written for this reader
- Governing the Promenaut Workforce · About 20 minutes
- Model Risk Evidence on Promenaut · About 30 minutes
The workforce and its governance, on the models you choose.
Frameworks, runtimes, automation and suite agents are other ways to get agents. Promenaut is the workforce itself, with the controls a regulated organisation needs, beside whatever you already run.
Promenaut · governed digital workforce
Modules that already do the work, and the ones your teams build
Approval where the workflow says, cost estimated before a run, a verdict from each run, in your own cloud account if you choose.
What is a governed digital workforce? →Other ways to get agents, beside it
Agent frameworks
The parts to build an agent’s logic.
Agent runtimes
Where an agent you built is hosted.
Automation
Fixed flows across apps, with AI as a step.
Suite agents
Agents inside one vendor’s suite, over its data.
Models · underneath
Anthropic, OpenAI and Google directly, Amazon Bedrock, or your own AI gateway
Chosen per agent, with a fallback that can cross providers.
One run, from brief to decision.
A requirements workflow, replayed: agent teams work in turn, artifacts accumulate, and the run waits for a person at each gate.
Full Requirements Intelligence
Readiness verdict
0 of 6 artifactsweighing signals…
See it on a workflow you own.
Bring a real process. We’ll show what your first agentic workflow looks like on the platform.
