AI Readiness Partner

Pillar 2 of 6

AI does not fix broken processes.

A process is not fine because there is a Notion page describing it. AI does not smooth out the gap between the documented process and the actual one. It runs the gap at scale. This page lays out what process readiness actually looks like, and how to spot the gaps before AI does.

Every business has processes. Almost no business has the processes it thinks it has. Org charts get drawn, SOPs get written, policies get filed in a shared drive nobody opens. The actual work runs along well-worn human shortcuts that exist outside any documented flow. AI does not learn those shortcuts. It erases them, formalises the wrong path, and runs it a thousand times before anyone notices.

What process readiness actually means

Process readiness is not whether you have written down what you do. It is whether the written version matches the lived version, and whether what gets lived can survive being run faster, more often, by a machine that does not improvise.

The gap between those two versions of the same process is where most AI investments get spent. Closing it is what our process optimisation engagement is built to do. The work depends on the disciplines layered into your people and capability pillar; without that, no documentation tool holds.

Documentation

Is the actual current-state process written down end to end, by someone who could hand it over without a six-month shadowing period? Most teams have aspirational documentation. The work runs on the difference between what is written and what is done, and only the people doing it know what that difference is.

Consistency

Does the process run the same way every time, or does it depend on who is holding the wheel on a given day? AI assumes the documented path is the real path. If three senior people each run the same workflow slightly differently, the model will pick one, formalise it, and lose the institutional intelligence in the other two.

Measurement

Do you know how long each step takes, where it breaks, and what your rework rate is? Without baseline numbers, you cannot tell whether AI has improved anything. You will get faster outputs. You will not know whether they are better, cheaper, or quietly creating new problems downstream.

Decision points

Are the choices made within the process rules-based and explicit, or opinion-based and invisible? Every business process is a chain of decisions disguised as steps. If a decision rule lives only in someone's head, AI cannot replicate it. It can only replace it with a guess.

Handoffs

When work passes between people, teams, or systems, does anything get lost, delayed, or duplicated? Handoffs are where AI looks most attractive on a deck and works least well in practice. They concentrate the messiness no other dimension fully captures, and they are the first place a model will trip.

The four readiness bands for Process

Our audit scores your process readiness from 5 to 20 across five questions. The bands tell you whether you are looking at faster automation, slower fixes, or a fundamental rebuild before AI sits on top.

  • Ready
    17–20

    Foundations are sound. Automate with intent.

    Core processes are documented end to end and the documented version matches what people actually do. Decision rules are explicit. You measure cycle time, rework, and exception rate. Handoffs are clean. At this band, AI accelerates outcomes that were already working. The risk now is automating something well that did not need to exist.

  • Progressing
    12–16

    Documented mostly. Lived occasionally.

    Most teams sit here. Your top three processes have written versions, but the gap between written and lived is wide enough to matter. Measurement is patchy. Exceptions get treated as anomalies when they are most of the work. The job is to close the documentation-to-reality gap one process at a time, starting with the one AI would land on first.

  • Developing
    7–11

    AI will surface every gap you have not closed.

    Some processes are documented. Many are tribal knowledge. The current-state map and the actual work disagree on more than half the steps in any flow you would care to automate. At this band, layering AI on top accelerates failures faster than fixes. Spend on the groundwork. Then come back to AI.

  • Critical
    5–6

    Stop. Map the work before you automate the chaos.

    No serious process documentation. Each person has their own way. Exceptions are the rule. Handoffs are verbal, undocumented, and dependent on a small number of people staying employed. Putting AI on this is not a project, it is a press release. Map the work. Find the patterns. Then talk about AI.

Why most teams get this wrong

After running this audit across more than a hundred organisations, the same four patterns repeat. Each is human, not technical. Each is fixable. Each kills more AI investment than any model limitation we have ever seen.

They confuse "we have an SOP" with "the work actually runs like the SOP says"

The SOP is what people would do if they had time. The actual process is what they do because they do not. We have audited businesses where an official seven-step onboarding ran in fourteen steps every time, and nobody had updated the seven-step version since 2021. The model will not read the truth. It will read the document.

They map the process they wish they had, not the one they actually run

Process mapping workshops have a known failure mode. The senior people in the room describe the ideal flow. The junior people who actually run it nod politely. The document that comes out is a fiction everyone has agreed to. We sat through one where the COO described a marketing approval as a two-day cycle, while the marketing manager in the same room knew it took eleven days and routinely depended on a WhatsApp message to one specific person to move at all. AI built on that fiction does not surface the truth. It just makes the gap more expensive.

They treat exceptions as one-offs when they are actually the majority of the work

In most operational processes, the documented path covers perhaps 60 percent of cases. The remaining 40 percent is exceptions, handled by judgement, custom routing, or someone in ops who has been there long enough to know the unwritten rules. One business we audited had a "standard" onboarding that took three days for some accounts and five weeks for others, and could not explain the variance. The five-week accounts were not exceptions. They were the work.

They think process is a documentation problem, not a leadership problem

Documenting a process is the easy part. Holding the team to running it consistently, measuring it honestly, and updating it when the work changes: that is the discipline that breaks down. Combine weak process with the data quality issues almost every business already has, and the result amplifies at scale. No documentation tool fixes a culture that does not actually use process.

AI does not fix broken processes. It automates them.

What good actually looks like

"Process ready" is not the same as "process perfect". It is the discipline of knowing what you actually do, measuring whether it is working, and updating the description when reality moves. Most organisations that get this right are not the ones with the prettiest process diagrams. They are the ones whose diagrams match the work.

26%

of organisations successfully transition AI from proof-of-concept to production. The BCG figure is not a story about AI being immature. It is a story about pilots running on processes that were never going to survive being scaled.

Source: BCG, 2024

The bar is roughly this. You can name your core processes and describe what each one actually does in current-state, not aspirational, terms. You measure cycle time, rework, and exception rate on each. Decisions inside the process are written rules, not personal preferences. When someone leaves, the process keeps running. The technology pillar assumes you have got here; without it, any automation tooling will inherit the gaps you have not closed.

None of that sells AI seats. But it is the difference between intelligent automation that compounds and intelligent automation that produces an expensive audit trail of the wrong work. The Deep Dive report walks through your specific process gaps and the order to close them.

How the audit measures your process readiness

The Process pillar in our audit is five questions, each scored 1 to 4: documentation, consistency, measurement, decision points, handoffs. The total places you in one of the four bands above. The questions are deliberately blunt. Process readiness is one of the pillars most often softened by self-assessment, so the audit pushes back.

The free 7-minute version gives you the band, the score, and a teaser of where your biggest process gap sits. The £97 Full Report breaks down each pillar in writing. The £497 Deep Dive takes that further with a 30 / 60 / 90 day plan personalised to your sector and operating model. If you would rather talk to a consultant about process specifically, we also offer process optimisation as a standalone engagement.

Take the free 7-minute assessment No card. No sales call. Just your score across all six pillars.

Frequently asked questions

What does "process-ready for AI" actually mean?

Process-ready means the documented version of how work gets done matches the lived version, decisions are explicit, measurement is in place, and handoffs are clean. It is not whether you have process documentation. It is whether what is documented runs the same way every time. Most organisations have process-ready pockets, not process-ready foundations.

Why does AI fail on processes that look fine on paper?

Because AI does not run on paper. It runs on what people actually do. Documented processes are usually aspirational. The lived version is messier, with shortcuts, exceptions, and unwritten decision rules that nobody flags as part of the process. AI built on the document misses the lived reality and surfaces the gap at scale.

Do I need to document every process before using AI?

No, but you do need to know which processes AI will touch and have those documented to current-state, not aspirational, accuracy. Start with the three that matter most for your use case. Map them as they are run today, including the exceptions and decision rules nobody has written down. If you cannot do that, those processes are not ready.

What is the difference between a workflow and a process?

A workflow is the technical sequence of steps a system or person executes. A process is the broader context: the goals, decisions, ownership, exceptions, and measurement around the workflow. AI tools are good at automating workflows. They are useless at fixing processes. Confusing the two is one of the most expensive misreadings in AI investment.

Can AI work with informal or ad-hoc processes?

Briefly, badly, and at scale. AI can be pointed at ad-hoc work, but it cannot improvise. It will pick a path through the ambiguity and formalise it, which is fine if you wanted that path and a disaster if you did not. The honest answer is that ad-hoc processes need to be made explicit before AI is useful, not the other way around.

How do you measure process readiness?

Five dimensions, each scored 1 to 4: documentation, consistency, measurement, decision points, handoffs. The total places you in one of four bands. The free seven-minute version of our audit returns the score and your biggest gap. The paid reports walk through each dimension in detail and give you the order in which to close gaps.

Where do most companies fail on process readiness?

Consistency and measurement. Documentation gets attention because it produces a deliverable. Consistency does not, and measurement requires baselines most teams do not have. The gap between documented and lived process is invisible until AI lands on top, and then it is visible in production, which is the worst place to discover it.

Is process optimisation the same as automation?

No. Optimisation is improving how the process runs. Automation is running it faster with less human input. Automation without optimisation makes a bad process faster and more expensive. Most consultancies sell automation as optimisation because automation is what the tools do. We do not. The order matters.

See where your processes actually stand.

Seven minutes. Thirty questions. A scored band for Process and the five other pillars that decide whether AI works in your business.

No card. No sales call. Just your score across all six pillars.