---
title: Workflow Intelligence as a Service (WIaaS)
description: WIaaS is a new infrastructure category — distinct from orchestration (LangGraph, CrewAI) and observability (LangSmith, Helicone). WIaaS operates before execution, scoring and shaping workflows deterministically with zero LLM calls.
canonical: https://cromus.ai/workflow-intelligence-as-a-service
source_html: https://cromus.ai/workflow-intelligence-as-a-service
---

# Workflow Intelligence as a Service (WIaaS)

> A new infrastructure category for AI workflows, operating **before execution**.

WIaaS sits between workflow design and workflow execution. It analyses, scores, optimises, and governs AI workflows **before** they consume compute resources, using deterministic intelligence and zero LLM calls.

Cromus is the first WIaaS platform and the company that defines the category.

---

## Where WIaaS sits in the stack

| Category | Operates | Examples |
|----------|----------|----------|
| **Workflow Intelligence (WIaaS)** | **Before execution** | **Cromus** |
| Orchestration | During execution | LangGraph, CrewAI, AutoGen, n8n |
| Runtime governance | During execution | Anthropic Claude Managed Agents, Google Agent Platform, AWS Bedrock Agents, Azure |
| Observability | After execution | LangSmith, Helicone, Langfuse |
| Workflow automation | During execution | Zapier, Make, n8n |

WIaaS is the only category that operates entirely upstream. It produces the governed, costed, validated input that the other categories execute and observe.

---

## What a WIaaS platform does

1. **Compile** — turn SOPs into portable SKILL.md files
2. **Validate** — check SKILL.md, ETHOS.md, MEMORY.md against open specs
3. **Score** — assign a deterministic Croms™ waste score
4. **Simulate** — project per-run and monthly cost across 5 cost-quality modes and 62 verified models
5. **Optimise** — surface concrete actions (downgrade model, parallelise, dedupe context)
6. **Export** — emit portable artifacts for LangGraph, CrewAI, n8n, AutoGen, Ollama, Claude Code, Replit AI, Open WebUI

All deterministic. All before a single token is spent.

---

## Why "as a service"

WIaaS is delivered through:

- **Hosted web app** — interactive Skill Studio, cost simulator, workflow canvas
- **MCP server** at `https://mcp.cromus.ai/mcp` — six deterministic tools, zero LLM calls, no-retention contract, callable from Claude Cowork, OpenAI Codex CLI, and any MCP-compatible client
- **Open specs** — SKILL.md, ETHOS.md, MEMORY.md, each with a free public validator
- **Skill packs** — pre-built workflow templates, both in-app (14 sample packs) and downloadable (10 startup ZIP packs)

---

## Distinct from adjacent categories

- **Not orchestration.** Cromus doesn't run your workflow; it compiles a portable spec that any orchestrator can run.
- **Not observability.** Cromus never sees a production token; it scores the workflow before it executes.
- **Not a model gateway.** Cromus doesn't proxy traffic; it recommends which model channel a workspace should use.
- **Not runtime governance.** Cromus is upstream of runtime stacks; it produces the input they consume.

---

## Related concepts

- [Pre-Execution Governance](/pre-execution-governance.md)
- [Croms™](/croms.md)
- [Claude Managed Agents Governance](/claude-managed-agents.md)
- [SKILL.md Validator](/validator/skill.md)