Spec Examples Physical AI Runtime Library Convert View Guide (Expert) Easy Guide GitHub Download WIA SOOM
.wia

The AI-Native Document Format for Physical AI

Humans write. AI executes. Machines verify. Robots converge.

A Markdown superset — every .md is a valid .wia. Now with ROS 2, fleet ops, and desired-state convergence.

v1.1 · Open Standard · Apache-2.0 / CC-BY-4.0 · Now with Physical AI & ROS 2

Four Properties.
One Format.

Existing formats cover one or two of these. Markdown is readable but not executable. Terraform is executable but not human-friendly. AGENTS.md is AI-aware but not verifiable. .wia unifies all four.

Human-
Readable
(Markdown)
Machine-
Executable
(Terraform)
AI-
Native
(AGENTS.md)
Self-
Verifying
(.wia)
.wia

From Manual Steps to Auto-Execution

The same runbook. Radically different workflow.

Before (.md)
1.Open wiki
2.Find the right page
3.Copy command
4.Paste in terminal
5.Check output manually
6.Screenshot for audit
After (.wia)
1.Open .wia file
2.Click "Run All"
3. Auto-executed
4. Auto-verified
5. PASS / FAIL badge
6. Execution log = evidence

Compact by Design

Everything you need. Nothing you don't.

6
Block Types
sh · sql · ai · verify · display · desired-state
14
Block Attributes
run · on= · expect= · model= · timeout=
retry= · watch= · rollout= · drift= · converge=
7
Input Types
string · number · boolean
enum · secret · file · array
12
Special Variables
{{$prev}} · {{$timestamp}} · {{$hostname}}
{{$memberId}} · {{$memberType}} · {{$memberRole}}

Built for the AI Era

Six capabilities that set .wia apart from every other document format.

Executable Blocks

sh run, sql run on=, ai, verify expect= — run code blocks in any language with declarative attributes.

Self-Verification

PASS/FAIL badges from declarative assertions. Every execution is self-validating — no external test harness needed.

AI Integration

First-class ai blocks with model selection, vision support, and Soomy auto-agent orchestration.

Safety Guards

Approval gates, dry-run mode, automatic rollback, and idempotency checks — production-safe by default.

Real-time Monitoring

watch=5s for continuous monitoring. Fleet-wide execution across hundreds of servers.

Compliance Ready

Execution log equals audit evidence. Export to HTML for SOC 2, ISO 27001, and regulatory compliance.

🤖

Physical AI Native

ROS 2 config drift, SROS2 security, URDF extraction, bag analysis, fleet canary deploy, and OTA orchestration — built into the format.

🚀

Desired-State Convergence

Declare the desired state. .wia auto-converges — the Terraform/ArgoCD for robot fleets. No equivalent exists elsewhere.

See .wia in Action

A real server health check runbook — readable, executable, and self-verifying.

server-health-check.wia
---
wia: "1"
title: Server Health Check
agent: soomy
inputs:
  server: { type: string, default: "prod-1" }
---

## 1. Check Disk Usage
sh run
df -h / | tail -1 | awk '{print $5}'
42%
verify expect="< 80%" ✓ PASS
Disk usage is within safe limits.
ai model=soomy
# Analyze the disk usage and suggest cleanup if > 60%
Summarize disk health for {{server}}.
Disk usage on prod-1 is 42%, well within safe limits. No action needed. Next check recommended in 24h.

ROS 2 Fleet Diagnostic

A real robot fleet health check — with config drift detection, canary deployment, and AI analysis.

ros-fleet-diagnostic.wia
---
wia: "1.1"
title: ROS 2 Fleet Diagnostic & Recovery
agent: soomy
inputs:
  namespace: { type: string, default: "/" }
  rollout: { type: enum, of: ["10,50,100", "25,100", "100"] }
---

## 1. Config Drift Detection
sh run drift=detect 0 drifts
ros2 param get /nav2 use_sim_time  # expected: false
ros2 param get /slam max_range     # expected: 12.0
verify expect=0 ✓ PASS
All parameters match desired state.
## 2. Canary OTA Update
sh run rollout={{rollout}} approve
sudo apt-get install -y ros-humble-nav2-bringup
# Rolls out 10% → 50% → 100% with health gates
# Auto-rollback if failure rate > 30%
Stage 1 (10%): 2/2 ✓ | Stage 2 (50%): 10/10 ✓ | Stage 3 (100%): 20/20 ✓
ai model=soomy
Analyze fleet health after OTA.
Check SROS2 security posture.
Generate SOC 2 compliance evidence.
Fleet health: 20/20 online, avg CPU 34%, 0 config drifts. SROS2 enabled on 18/20 (2 pending). SOC 2 CC7.2 monitoring: Compliant. Risk score: 12/100 (Low).

.wia vs. Alternatives

How .wia stacks up against existing formats and tools.

.wia Ansible Terraform InOrbit Foxglove AGENTS.md
Human-readable ~
Executable ~
AI-native ~
Self-verifying ~ ~
ROS 2 / Physical AI
Config drift detection ~
Canary fleet deployment
OTA orchestration ~ ~
Compliance reporting ~
Open standard ~

Where .wia Shines

From incident response to onboarding — any process that needs to be documented, executed, and verified.

🛠

DevOps Runbooks

Incident response, deployment checklists, and rollback procedures. Click "Run All" instead of copy-pasting commands at 3 AM.

🤖

Physical AI & ROS 2

Config drift detection, SROS2 security automation, URDF extraction, and ROS bag AI analysis. The .wia for every robot in your fleet.

🚀

Fleet Canary Deployments

OTA updates with 10%→50%→100% staged rollout. Health-gated auto-rollback. No more all-or-nothing firmware pushes.

🔒

Robot Compliance

SOC 2, ISO 27001, and ISO 9001 compliance reports for robot fleets. SROS2 posture assessment. The runbook is the audit evidence.

🎓

Desired-State GitOps

Declare the desired state of your robot — parameters, nodes, configs. .wia converges automatically. The Terraform for Physical AI.

$15.28B

Physical AI Operations Market

The Physical AI market is projected to reach $15.28B by 2032 (CAGR 47.2%). Robotics VC funding hit $27.6B in 2025 alone. Yet there is no standard format for robot operations — no docker-compose.yaml equivalent for robot fleets. .wia fills this gap: config drift detection, canary deployments, OTA orchestration, and compliance reporting — all in one declarative, AI-native format.

Learn. Certify. Operate.

WIA Standards publishes the world's most comprehensive Physical AI guidebook series.
Each guide pairs with .wia runbooks for hands-on practice.

🤖

AI Embodiment Standard

Physical AI implementation standards — sensors, actuators, embodied intelligence, safety certification.

EN → KO →
🧠

Brain-Computer Interface

BCI standards for neural signal processing, EEG interfaces, and assistive technology integration.

EN → KO →
🙌

Emotion AI Standard

Affective computing, facial/voice emotion recognition, ethical AI guidelines.

EN → KO →

Smart Wheelchair Standard

Autonomous mobility, obstacle avoidance, accessibility-first design for powered wheelchairs.

EN → KO →
Generative AI vs Physical AI — Full Book →

Available in Korean & English at wiabook.com

Everything You Need

An open standard backed by production-ready tooling.

💻

WIA SOOM

Reference runtime. AI-First Cloud Operations Platform for Windows, macOS, and Linux.

Download →
📜

Spec v1

Full specification: block types, attributes, frontmatter, variables, and execution model.

View on GitHub →
📖

111 Example Runbooks

Production-ready templates across 18 categories — DevOps, security, onboarding, and fleet operations.

Browse Examples →
🌐

254 Languages

WIA SOOM supports 254 languages out of the box. Global teams, local experience.

Learn More →