AI Readiness Assessment
We check whether your business is ready for AI before you spend money on it.
What is AI readiness assessment?
An AI readiness assessment is a structured review of your data, systems and working practices that establishes which tasks can be automated now, which need groundwork first, and which should be left alone. It produces a written verdict, not a sales proposal.
We look at three things. Where your data is kept and what state it is in. Which systems you use and whether they can be connected. How work actually gets done day to day, as opposed to how the manual says it is done.
Most problems we find are not AI problems. Records sit in spreadsheets, two systems do not talk to each other, or a process exists only in one person head. Automating any of that as it stands makes the mess run faster.
We interview the people doing the work, not only managers, because the documented process and the real one usually differ. That gap is where the time goes and where automation either pays or fails.
What we are actually paid for
Duara runs readiness assessments from Nairobi for businesses across Kenya. Our assessors spend most of their time in operations rooms and back offices rather than boardrooms, because that is where the evidence is.
We are paid for the assessment, not for what it recommends. That matters, because it means a finding of "do nothing yet" costs us nothing to write. Ring 0723658888 if you want to test that claim before booking.
Who this is for
- You have been pitched AI but cannot judge the claims
- Your data sits in systems that do not connect
- You want a clear answer before committing budget
- A previous AI project stalled and nobody knows why
Industries we serve
- Banking and microfinance
- Healthcare and clinics
- Retail and eCommerce
- Logistics and transport
- Manufacturing
- Professional services
Problems this solves
Businesses usually book an assessment after one of these four situations, and often after more than one at once.
- Vendor claims you cannot verify. Suppliers promise savings without showing what has to be true for those savings to appear, and you have no independent way to check.
- Data spread across disconnected systems. Records live in a mix of spreadsheets, accounting software and paper, so nothing can see the whole picture at once.
- Undocumented processes. The real process differs from the written one, so anything built against the documentation fails on contact with the actual work.
- Budget committed before scope is understood. Money is approved on a proposal that was never tested against your data, your systems or your staff capacity.
What we deliver
A written assessment that ranks your tasks by automation potential, states the groundwork required, and gives you a costed view before any build is agreed.
A written inventory of your systems, data sources and how they connect
Your tasks ranked by automation potential against the effort required
The groundwork that must happen before any AI work begins
A clear statement of what we would not automate, and why
Estimated build cost and running cost for the work we do recommend
A report written so your board or finance team can question it
Features
- Interviews with the people doing the work. Not only management, because the process on paper and the process in practice are rarely the same thing.
- Data quality measurement. We check whether your records are complete, consistent and free enough of duplicates for anything to be built on them.
- System connection check. We establish which of your systems expose their data and which would need custom integration work first.
- Task ranking by effort and return. Every candidate is scored, so the sequence of work rests on evidence rather than on whoever argued hardest.
- An explicit do-not-automate list. We name the tasks where automation would cost more than it saves, so nobody revisits them in six months.
Benefits
- You avoid spending on the wrong thing. The cheapest project of all is the one you correctly decide not to run.
- Finance gets a number it can interrogate. Assumptions are written down, so your accountant can argue with them instead of accepting a supplier figure.
- Groundwork surfaces early. Missing integrations and dirty data appear before a build starts rather than halfway through one.
- Staff are involved from the beginning. People who were interviewed during the assessment are far more likely to use whatever gets built.
Technologies we use
- Your existing systems. We assess the CRM, ERP, accounting package and spreadsheets you already run, rather than assuming you will replace them.
- Data profiling. Automated measurement of completeness, duplication and consistency across the records you hold.
- Process mapping. Documented flows showing each step, who performs it, how long it takes, and where work sits waiting.
- Cost modelling. Total cost of ownership including licence fees, usage charges and maintenance, not only the build price.
- M-Pesa transaction review. Where reconciliation is in scope, we assess the mobile money records directly rather than working from summaries.
An assessment uses inspection tools rather than build tools. Nothing is installed permanently and nothing changes in your systems.
How the process works, step by step
An assessment runs over two to three weeks. The first call is free and there is no obligation at any stage.
-
Step 1
Free scoping call
We agree which departments and systems are in scope and what a useful answer would look like for you.
-
Step 2
Evidence gathering
We interview the people doing the work and take copies of the records we need to profile.
-
Step 3
Data and system profiling
We measure data quality and establish which systems can realistically be connected.
-
Step 4
Findings written up
You receive the ranked task list, the groundwork required, the costs, and what we advise against.
-
Step 5
Walkthrough with your team
We present the findings and answer challenges from your managers and finance team.
Quality assurance and how we work
What you can hold us to on an assessment specifically.
- We assess, we do not sell. The fee covers the assessment. A recommendation to do nothing costs you the same as a recommendation to build.
- Every claim traces to evidence. Findings reference the interview, the record or the measurement they came from, so you can check them.
- Assumptions are written down. Where we estimate rather than measure, we say so and state the confidence level.
- Your data is not copied off-site without agreement. Profiling happens on your systems or on extracts you approve, and copies are destroyed afterwards.
- The report stands alone. It is written so another supplier or your own team could act on it without us.
Frequently asked questions about AI readiness assessment
What is an AI readiness assessment?
A structured review of your data, systems and processes that establishes what AI can realistically do for your business now, what needs fixing first, and what is not worth automating at all.
How long does a readiness review take?
Two to three weeks for a small or medium business, including the interviews. Organisations with many systems or several sites take longer.
What does a readiness review cost?
It depends on how many departments and systems are in scope. We quote a fixed figure after the free scoping call.
What if the finding is that we are not ready?
Then we say so and set out what to fix first. Discovering it now costs a fraction of discovering it halfway through a build.
Do we need technical staff to take part?
No. We work with owners, managers and the people doing the work. The technical interpretation is ours to do.
Who exactly do you interview?
Whoever performs the work daily, plus whoever owns the process. Management alone gives a picture that is usually too tidy.
Will the assessment disrupt operations?
Barely. Interviews are short and booked around your work, and we do not need to pause any system.
What documents do we receive?
A system and data inventory, a ranked task list, the required groundwork, costs, and an explicit do-not-automate list.
Can our board challenge the findings?
That is the point of writing assumptions down. Directors and finance can argue with the reasoning rather than the conclusion.
Do you recommend your own build work?
Where a build is justified we say so and state plainly that we are a candidate for it. You are free to take the report elsewhere.
Is the report useful if we never hire you?
Yes. It is deliberately written so your own team or another supplier can act on it without us involved.
Our records are messy. Is that a problem?
Most records are. We measure how messy and tell you what has to be cleaned before anything can be built on top.
Do you assess systems we plan to replace?
We take planned replacements into account, because building against a system you are about to retire wastes the budget.
How are tasks ranked?
By volume, how rule-driven the task is, what it costs you today, and how much effort automating it would take.
What do you find most often?
That the real bottleneck is a missing integration or an undocumented process, neither of which needs AI to fix.
Can you assess a single department?
Yes, and it is often the sensible scope. Finance, customer service and HR are the usual starting points.
Is what you see kept confidential?
Yes. Records stay with us under agreement, are never used to train public AI models, and extracts are destroyed after the work.
What happens once the assessment is done?
You decide. If you proceed, the ranked list becomes the roadmap. If you do not, you still know exactly where you stand.
How is this different from a supplier proposal?
A proposal argues for work. An assessment reports what is true, including when the honest answer is that you should do nothing yet.
Can you reassess later?
Yes, and businesses often do after fixing groundwork. A second assessment is faster because the inventory already exists.
Glossary of terms
Terms used in readiness work, in plain language.
- AI readiness
- Whether your data, systems and processes can support automation today, without further groundwork.
- Baseline
- A recorded measure of how long a task takes now, used later to check whether anything actually improved.
- Data quality
- How complete, consistent and duplicate-free your records are. Poor data quality is the most common blocker we find.
- Data profiling
- Automated inspection of records to measure how complete and consistent they are.
- Integration
- A connection letting two systems exchange data automatically instead of someone retyping it.
- Total cost of ownership
- Build cost plus everything it costs to keep running, including licences, usage charges and maintenance.
Related services in AI Consulting
Talk to us about AI readiness assessment
Book the free scoping call and we will tell you whether an assessment is even necessary. Sometimes a single conversation is enough.
Start your project