Public Sector27 August 20266 min read

No AI Without Information Architecture — Part 2 of 10

Start with one form

We used to sell big systems to people who had no idea what we were talking about. Now we ask for one paper form. It is a smaller sale to the same destination — and it actually closes.

#Public Sector#Digital Transformation#Delivery#Solution Architecture

For years the approach was to sell the platform. Turn up with the architecture, explain the warehouse, describe the analytics layer, sketch the AI roadmap. It is a genuinely good proposal. It is also, in retrospect, obvious why it kept not closing: you could watch across the table as people stopped following, and nobody in the room could approve a thing they could not picture.

What changed everything was asking a much smaller question. Give us the form your field workers fill in by hand. We will digitise that one.

Two ways to open the same conversationPROPOSAL A — THE WHOLE PLATFORMData warehouse, lakehouse, governance framework, analytics layer, AI roadmapno decision, no budgetPROPOSAL B — ONE FORMGive us the form your field workers fill in by hand. We will digitise that.understood, approvedProposal B is a smaller sale and the same destination
The first is a bigger, more valuable, more accurate proposal. It is also the one that does not get bought, because nobody in the room can picture it or approve it.

Why the smaller ask wins

Both proposals lead to the same place. The difference is entirely in what the person opposite has to do with them.

The platform proposal asks them to imagine an end state, believe a multi-year benefit, find budget that does not exist in this cycle, and defend all of it to people who were not in the room. That is a lot to ask of someone who met you an hour ago.

The one-form proposal asks them to hand over something they already have, which is already causing them irritation, in exchange for something they can see working in weeks. There is no imagination required. They know exactly what that form is because they have seen it come back illegible, late, and in a pile.

Everything after that gets easier, because you are no longer a supplier with a proposal. You are the people who fixed the form.

The form was never the product

Being clear about this internally matters, because it is easy for a team to start believing the small thing is the whole engagement.

What one form is really forOne paper formdigitisedStructured dataarriving weeklyA view nobodyhad beforeThey ask forthe next thingscope widens because they widened itScope creep, when the client is the one doing the creeping, is the signal you were hoping forThe alternative is a platform proposal that is more correct, more complete,and still sitting unread
The form is not the product and was never meant to be. It is the cheapest possible way to start producing structured data about something the organisation already cares about.

Digitising one form does something specific and disproportionately valuable: it starts producing structured data, weekly, about something the organisation already cares about. Not a data lake. Not an integration programme. A steady trickle of clean records where previously there was a pile of paper.

And then — this is the part that reliably happens — someone looks at the resulting view and sees something they did not know. Not because the analysis was clever, but because nobody had ever been able to look at that operation in aggregate before.

That is the moment the conversation changes. They ask what else it could show. They ask whether it could cover the neighbouring region. They ask whether the supervisors, who are already out there anyway, could record two more things while they are at it.

Scope creep, when the client is doing the creeping, is the signal you were hoping for. It means the value landed. It also means the expansion is being pulled rather than pushed, which is a completely different commercial position to be in.

Choosing which form

Not every process is a good wedge. What we look for:

It already hurts. Not theoretically, weekly. Someone is chasing these forms and complaining about it.

It has breadth. Some processes touch a narrow slice of the organisation; others reach almost everywhere. A field programme that already has staff in every district is worth far more as a collection point than a bigger process confined to head office — you are not just digitising a form, you are choosing a sensor network. Given two candidates, take the one with the wider footprint even if it is the less glamorous process.

Its data leads somewhere. You should be able to say, before you start, which decision better data about this would improve. If you cannot, you are about to do work that will be appreciated and then stop — which is the subject of a later article in this series.

It is bounded. One form. Not the form and its four dependencies and the system it feeds.

The trap on the other side

There is an honest risk in this approach and it is worth stating plainly, because the approach is easy to over-apply.

If you only ever digitise forms, you become a forms supplier. The work stays small, the relationship stays transactional, and someone else eventually turns up to do the platform. The wedge only works if you actually drive it — if the first delivery is deliberately followed by a conversation about the foundation underneath, at the moment the client is most receptive to having it.

The other trap: governance cannot wait for phase two, even though it is not the opening message. The form you build in week three will be collecting real information about real people, and the decisions you make then about who can see it, how long it is kept, and who owns it are decisions whether or not you make them deliberately. Being quiet about governance in the sales conversation is a positioning choice. Being quiet about it in the build is negligence.

What it looks like when it works

The pattern we have now watched several times: one bounded process digitised, structured data starting to accumulate, a simple view produced from it, and then the client themselves widening the scope — first to more regions, then to more categories, then to things adjacent to the original process entirely.

At which point you are no longer selling a data platform. You are being asked what it would take to support what they have already decided they want. That is the same conversation you tried to open with. It just arrives with a sponsor attached.

How CloudNala can help

The judgement in this approach is choosing the right first process — one that hurts enough to get approved, reaches wide enough to be worth having, and leads to a decision somebody actually makes. We spend most of the early effort there rather than on the build itself, because picking the wrong form produces a successful little project that goes nowhere.


Work with CloudNala

CloudNala helps organisations move from technology ambition to practical execution across cloud, AI, data, platform engineering and digital services.

Whether you are exploring AI, modernising your cloud environment, building a public-sector digital service, or turning an idea into a working MVP, we can help you shape the roadmap and deliver the next step.

Book an AI Readiness Workshop or write to us at consult@cloudnala.co.za