Document 02 · Method

How we operate.

7 levels of AI-operation maturity journey — together with the community.

4,100 words · 18 min read Companion to Thesis v10 Cloudscockpit.io · v10.0
Token cost · L1 → L7
89%
A mature AIOps operator cuts token cost by 89% vs. baseline LLM spend.
Compounding execution rate lift
+80%
Execute more actions with fewer tokens — our guidance patterns your focused goal.
Community-first
$0
We drop the cost of experimentation to zero for communities facing job loss and AI risk.
Section 01 · The System 01

How Actionboard works.

Four operation layers stacked on top of each other. A human pilot at the top, the Voltron AIOps five Lion-core agents running in formation underneath, the RAOARA loop running through them, and an Action Graph capturing everything they do. Every cycle the operator finishes writes a permanent node — the system gets smarter at this specific operator's work without ever leaving the operator's hands.

Layer 01
Pilot
P
Human · Pilot
Decides, consents, owns the outcome. The agent does not execute without the human. Permission is a first-class type, written to an auditable ledger.
↓ pilots the formation ↓
Layer 02
Formation
Black
Orchestrator
Planning, architecture, execution sequencing.
Green
Retriever
RAG knowledge fetch and context assembly.
Blue
Builder
Code gen, implementation, test scaffolds.
Yellow
Security Gate
Scans, vulnerabilities, IAM, compliance.
Red
Deployment
Git ops, release, rollback, runtime.
↓ runs on the loop ↓
Layer 03
Loop
R
Recognize
Goal
A
Acquire
Data
O
Organize
Structure
A
Apply
Action
R
Reflect
Proof
A
Amplify
Scale
↓ writes to the graph ↓
Layer 04
Memory
Action Graph
Directed acyclic graph of every recognized goal, acquired source, organized plan, applied action, reflected proof, and amplified pattern. Crystallised IP — belongs to the operator, not the vendor.
The cycle: the pilot states a goal, the formation acquires + organizes + applies, RAOARA reflects and amplifies — and every action becomes a permanent node in the graph. Every future loop starts smarter than the last.
Section 02 · The Adoption 02

Adoption climbs the AIOps ladder.

Actionboard isn't switched on at procurement — it's earned, one rung at a time. Each level turns a new capability online, retires a class of manual work, and unlocks the next tier of automation. Operators don't buy maturity; they pay for it in completed RAOARA loops.

The ladder is the same diagnostic frame we use to underwrite operators and to scope cohort engagements. Pick the rung you're on today; the column on the right names what comes online when you finish climbing to the next one.

Level Capability & what unlocks Automation Ops risk Token cost
L1
AI Action Apprentice
Manual execution · basic tool usage · single operator.
Unlocks: pilot a first workflow under supervision
85% base
L2
AI Action Practitioner
AI-assisted actions · auto-execution begins on safe paths.
Unlocks: Voltron Desktop · Black Lion orchestration
70% 85%
L3
AI Action Specialist
Action Graphs replayable · workflow automation across the team.
Unlocks: Action Graph replay · guided simulation
55% 65%
L4
AI Action Lead
Full five-lion formation · cross-pod orchestration online.
Unlocks: Voltron Agent Server · multi-team pods
38% 48%
L5
AI Action Architect
End-to-end pipelines · enterprise governance · CIO collaboration.
Unlocks: Pattern Registry · CIO invite · PRD gate
22% 30%
L6
AI Action Visionary
Near-full automation · pioneering pattern discovery in sim.
Unlocks: Fast Action Simulation · steered SLMs
12% 18%
L7
Chief AutoAction Officer
Autonomous RAOARA · Proof-of-Thought governance.
Unlocks: real-time pattern synthesis · self-validation
5% 11%
Capability at each rung is observable, measurable, and audited against three axes: depth of automation, operational risk, and per-action token cost. You climb by completing loops — not by accumulating licences.
Section 03 · The Map 03

The Action Graph, in detail.

Inside the system, every goal fans out across the five RAOARA columns and converges again at audit + amplify. The Voltron formation walks this graph. The Proof of Thought ledger reads from it. The graph is the single source of truth for what a team has done, why, and what to repeat.

Fig. 03 · Action graph topology RECOGNIZE ACQUIRE ORGANIZE APPLY REFLECT · AMPLIFY GOAL Scoped intent DOC SOURCE PDF · MD · DOCX API DATA REST · GraphQL WEB DATA RAG · scrape ACTION LIST ordered tasks STRUCTURED PLAN DAG segment API CALL DB UPDATE EMAIL DEPLOY QUALITY CHECK reflect AUDIT LOG proof of thought AUTOMATE amplify
Section 04 · The Engine 04

Pattern discovery, by simulation.

Higher AIOps maturity means exponentially faster discovery of optimal design patterns — through simulation, without touching production. Teams publish AIOps metrics and invite SMEs into the AI Labs ActionBoard to collaborate before any code moves to real-life users.

L1–L2 · Manual
3–4 weeks
Trial & error · no simulation · inconsistent results.
L3–L4 · Action Graph
3–5 days
Action Graph replay · guided simulation · reproducible.
L5–L6 · Fast Sim
4–8 hours
Fast Action Simulation · cross-pod validation in parallel.
L7 · Autonomous
Real-time
Autonomous pattern synthesis · self-validation loop.
L1–L2 · Manual
~700 hrs
L3–L4 · Action Graph
~96 hrs
L5–L6 · Fast Sim
~6 hrs
L7 · Autonomous
~30 min
0200400600800 hrs

Simulation → PRD pipeline

01
Run Simulation
Fast Action Sim · no PRD risk.
02
Validate Pattern
Action Graph + compliance.
03
Publish Metrics
AIOps scorecard · team visible.
04
Invite SMEs
Governance collaboration.
05
Deploy Pods
Cloud · Security · Ops · parallel.
06
PRD Approved
All pods green · pattern live.

Every gate backed by Proof-of-Thought audit. CIO reviews AIOps metrics at stage 4 before any production commit is allowed.

Section 05 · The Hub 05

Multi-team AI Labs Pods.

Each team deploys their own AI Labs Pod to build real-time KB updates, and action patterns are validated against goals in isolation. Patterns validate independently and in parallel — no team blocks another. All approvals converge at the PRD gate.

Fig. 05 · Labs Pod topology & PRD gate SA Architecture LABS POD Discovers & proposes patterns PROPOSE Pattern Registry AIOps metrics published & shared PROOF OF THOUGHT audit trail SME invite triggered here CIO Governance LABS POD Alignment · regulatory Cloud Infra LABS POD Deploy · scale · cost Security LABS POD Threat · IAM · gates Ops LABS POD Runbook · SLA · alerts PRD GATE all pods ✓
Section 06 · The Receipts 06

What the method has shipped.

Three years of customer experiments across NGO operations, garments factory operations, legal, HR, micro-tasking, DevOps, CloudOps, and UX use cases. Built using the ActionBoard agentic pattern — dedicated boards for Product Design, UI/UX, API Development, AI Engineering, MarketingOps, and LegalOps.

Throughput
80%+
More actions executed per operator running on the full RAOARA loop.
POC → Demand
100%
Every user that completed a Proof-of-Concept asked for the full version.
LLM Cost
70%+
Reduction via TRM-tuned Steered Small Language Models.
University Partners
7
AI labs in partnership with universities over the last year, including AIUB & six others.
NGO Ops Garments Legal HR Micro-Task DevOps CloudOps UX
Read the thesis →

Every complete cycle writes a node.
Every node compounds.
The graph crystallises.

— Cloudscockpit.io · Method · v10 · 05.14.26