AI readiness
Most organizations are a level below where they think they are.
AI is not one capability but five, and each one is built on the one below it. The gap between the level an organization is operating at and the level its roadmap assumes is the single most common reason AI investment returns nothing. Start by finding out which level you are actually on.
- Levels
- 5
- Self-check
- 6 questions
- Assessed properly
- A Jump Start
Start here
Which level are you actually on?
Answer six questions and find out. Nothing is captured and nothing is sent — the answer is yours whether or not you ever speak to us.
Six questions
No email, nothing sent anywhere, answer in the browser. The result is the level your answers actually support, not the one a roadmap assumes.
1Could you produce a clean, current dataset for a core business process this week — without a special project to assemble it?
2Is at least one model running in production, owned by a named team, with someone accountable when it degrades?
3Are generative tools embedded in how work actually gets done — in the toolchain and the product — rather than used individually by whoever chose to?
4Does anything complete a multi-step task end to end without a human in the loop, and are you comfortable that it does?
5Do multiple agents coordinate across systems on a real process, with the orchestration itself owned and measured?
6Can you state what your AI spend returned last quarter, against a target set before the work started?
0 of 6 answered
The ladder
You cannot skip a layer.
Each level is built on the one below it. That is not a maturity model invented to sell a service — it is a description of what breaks. Generative tools on ungoverned data produce confident nonsense. Agents on a fragmented toolchain automate the fragmentation. Orchestration without measurement cannot be told apart from activity.
Which is why the work of getting to the next level is almost never model work. It is data quality, ownership, delivery discipline and governance — the things we were already fixing before anyone called it AI readiness.
It is also why we do not treat the top of the ladder as the goal. The level worth reaching is the one your business goals actually require, and which rung that is depends on what you are trying to achieve — so that is the question we settle first, and the climbing we recommend is only ever as far as the answer demands.
Level 1: Machine Learning
Analyzes and predicts- Forecasting
- Fraud detection
- Churn prediction
- Automated pricing
Level 2: Neural Networks and Deep Learning
Recognizes patterns at scale- Computer-vision quality inspection
- Voice commands
- Document processing
Level 3: Generative AI
Creates content and code- Drafting
- Meeting notes
- Knowledge bases
- Code generation
Level 4: AI Agents
Executes multi-step tasks- IT tasks handled end to end
- Requests processed without a human in the loop
Level 5: Agentic AI
Orchestrates entire processes- Networks of agents collaborating
- Legacy modernization
- AI built into the product
You cannot skip the layers — each one builds on the last. Most organizations are stuck at level 1 or 2 while their competitors race to level 5.
Our own position on the ladder
We did not read about level 5. We shipped it.
The reason we can tell you where you are is that we have had to answer the question ourselves — with products in the market, not slides.
- Level 5
empwr.ai
An AI-native program assistant, incubated inside Envorso and spun out as an independent company in September 2024. It listens across meetings, workflows and engineering systems and turns unstructured conversation into structured outcomes — orchestration across systems, which is level 5 by definition. Over 100 companies in its first year, and SOC 2 Type 2 certified.
How a recurring client problem became empwr.ai - Level 5
Envorso Sports
An AI-first fan platform where the architecture came first and the AI followed from it — one fan identity rather than five vendor records, because personalization is impossible without it. Built from a live engagement rather than a market thesis.
From one franchise to a platform - Level 1
The software recall dataset
Every software-related vehicle recall since 1994, compiled and maintained rather than licensed. Level 1 done properly is unglamorous and it is the layer everyone skips — which is exactly why we publish ours.
1,565 recalls, 115.5M vehicles
Every one of those is the same argument we make to clients: the data layer comes first, the architecture decides what is possible later, and if you cannot measure the return you cannot tell capability from activity.
How you climb
Every practice moves you up a specific rung.
AI capability is not bought separately from delivery capability. Each of the four practices operates at particular levels, and this is which.
Strategic Portfolio Management
Portfolio work touches the whole ladder, because funding decides which layer gets built. It is also where level 5 measurement lives — an AI budget line assessed on outcomes rather than activity.
Read about this practiceProduct & Technology Delivery
Getting a model into production with a named owner is level 2. Embedding generative tools in the toolchain and the product is level 3. Shipping something that completes work unattended, safely, is level 4.
Read about this practiceCIO & IT Advisory
The estate decides whether level 1 is even reachable: data spread across applications nobody has counted cannot be joined, and no amount of model choice fixes that. At the other end, translating technology spend into business value is level 5 measurement — the same discipline, applied to the whole portfolio rather than one program.
Read about this practiceQuality & Operational Performance at Scale
Joining design, manufacturing and supply chain data is the level 1 work, and it is the layer most quality programs assume they already have. Pattern detection across warranty and test data is a model in production with an owner, which is level 2. Level 3 is where generative tooling starts earning its place in the analysis itself.
Read about this practiceHow we work on it
The same three steps, applied to the ladder.
AI readiness is not a separate engagement model. It is the existing one, pointed at a different question.
Evaluate
Which level are you actually at, versus the level your roadmap assumes? Is the data good enough to support the next one? Does the organization change fast enough to absorb it?
Improve
Build the layer that was skipped. Usually data quality, delivery discipline, or governance long before it is a model — which is precisely why the firms selling models alone keep failing.
Confirm
Eval-driven measurement. Outcomes, not activity metrics. Did the AI spend return anything, measured against targets agreed before the work started?
The claim
Every consultancy claims AI expertise. That makes the claim worthless.
So here is the version that can be checked, next to the version everybody says.
The claim everyone makes
We’ll help you adopt AI.
What we can prove
AI programs fail for organizational reasons, not technical ones — and organizational delivery is what we already fix.
The claim everyone makes
We have an AI practice.
What we can prove
We’ll tell you which level of AI capability you are actually at, versus the one your roadmap assumes.
The claim everyone makes
AI will transform your business.
What we can prove
AI won’t save you if you can’t change. Here is what has to be true first.
The claim everyone makes
We’ll build you an AI roadmap.
What we can prove
Your AI roadmap is already obsolete. Build capability, not a plan.
Find out where you actually are.
The self-check above gives you a reading in six questions. A Jump Start gives you the version that survives scrutiny — where you sit on the ladder, what is holding you there, and what has to be true to move.