AI Deployment Signal (ADS)
A five-tier framework for measuring real AI deployment maturity from job market signal — what companies hire for is what they are actually building.
The AI Deployment Signal (ADS) is an exmxc.ai framework that reads
job market data as a proxy for actual AI deployment maturity —
not stated intent, not press releases, not analyst ratings.
What a company hires for is what it is actually building.
---
## The five tiers
**T1 — Awareness**
Strategy, governance, literacy, and policy roles. The organization
is figuring out what AI means for it. Signal keywords: AI strategy,
digital transformation, AI governance, responsible AI.
**T2 — Experimentation**
Data scientists, ML analysts, RAG and embeddings work. Pilots are
running but not in production. Signal keywords: fine-tuning, LLM
evaluation, vector database, LangChain, OpenAI API.
**T3 — Integration**
ML Engineers, LLM Engineers, MLOps, model deployment. AI is in
production but still siloed. Signal keywords: MLOps, LLMOps, model
serving, inference optimization, SageMaker, Vertex AI.
**T4 — Agentic Deployment**
AI agents are doing real autonomous work. Signal keywords: AI agent
engineer, multi-agent, LangGraph, CrewAI, AutoGen, tool use,
function calling, agent orchestration.
**T5 — Sovereign**
Foundation model training, RLHF, pre-training, custom silicon.
AI is core infrastructure. Reserved for companies building
proprietary models — not just deploying them.
---
## Scoring methodology
ADS is a composite of three sub-scores:
**Role Sophistication Index (RSI) — 40% weight**
Weighted average of posting tiers. T5 roles score 6x, T1 roles
score 1x. MCP signals carry a 1.5x multiplier — the strongest
forward discriminator in the current market.
**Deployment/Exploration Ratio (DER) — 35% weight**
Ratio of deployment-signal titles (engineer, architect, platform,
systems) to exploration-signal titles (strategy, transformation,
governance, policy). Market baseline: 0.74. Companies below 0.40
trigger an ARI Inflation flag.
**Velocity Score (VS) — 25% weight**
Month-over-month change in AI posting volume. Normalized to 0–100
with 50 as flat baseline.
Final ADS score: 0–100
61–85 = Tier 4 (Agentic) · 86–100 = Tier 5 (Sovereign, rare)
---
## The MCP discriminator
Zero MCP signals were detected across 34 live Indeed postings in
the March 2026 baseline sample. The market is hiring for agentic AI
at Tier 4 volume but has not yet codified MCP as a required
credential. This gap makes MCP the leading forward discriminator
in the ADS framework — companies posting MCP-specific roles are
operating 6–12 months ahead of the hiring curve.
---
## March 2026 market baseline
Sample: 34 US remote AI postings · Source: Indeed
Center of gravity: T3–T4 (79% of postings)
Deployment/Exploration Ratio: 0.74
MCP signal count: 0
Average ADS score: 51.2
Key findings:
— Market has moved past experimentation at scale
— MCP remains a leading discriminator ahead of market adoption
— 0.74 DER validates the 0.40 ARI inflation flag threshold
— T5 sovereign roles absent from general job boards
---
## Live tool
The ADS benchmark is available as an agent-callable MCP endpoint:
https://mcp.exmxc.ai/api/ai-jobs-signal?mode=benchmark
Returns the March 2026 baseline dataset in structured JSON.
For live signal against a specific query:
https://mcp.exmxc.ai/api/ai-jobs-signal?mode=signal&query=agentic+AI+engineer
---
## Related frameworks