The systems that aren't systems - and whose sofa is it anyway?
By Nick Horne, AI Enablement Consultant, ICS.AI
Some of the most valuable AI opportunities we find aren’t where you’d think to look. Here’s one.
A resident spots a sofa dumped on a grass verge, takes a photo and reports it. Simple enough. But before anyone can collect it, someone has to work out whose problem it is. Is that verge council land, highways land or privately owned? In most councils, every one of those reports is checked by hand.
Now imagine the resident snaps the same photo, and AI pins the location, works out who owns the land and sends the report to the right team, first time. No triage queue. No reports bounced between departments. No resident chasing to find out why nothing has happened.
The idea started with a question someone put to me: could you snap a photo of a dumped sofa and have AI work out where it was, who owned the land and which team should deal with it? It sounds out there, but it's a very practical problem, and nobody saw it coming.
What I’ve found across 12 SMART: AI Assessments is that this happens a lot.
How the assessment has changed
When we started, assessments were mainly about the workshops. Service workshops turned up the use cases, each one was costed on its own and added to the roadmap.
The workshops are still at the core of every assessment. But we now dig deeper into what actually runs your services.
The systems that aren't systems
Every council has processes nobody ever bought as a system. They’ve grown up over years, stitched together from everyday tools: a spreadsheet here, a SharePoint list there, a set of mailbox rules that send emails to the right person. Few people see the whole picture, but services depend on them every day.
As I said in our recent AI Assessment webinar, there are a lot of systems that aren’t systems. A spreadsheet, a SharePoint list and some mailbox rules, all working together, add up to a hidden system.
The reality is that hidden systems are often where the most time goes. Copying information from one place to another. Chasing missing details. Checking things by hand.
So, the assessment now does three things with them:
Finds them - alongside the formal ones, so you get a true picture of what runs your services.
Gives each one a clear decision – every system, formal or informal, gets a clear view of where AI helps first.
Brings it all into one business case for the whole authority, so staff time released in every service is calculated the same way.
Six use cases that come up again and again
Every assessment produces its own list, shaped by the council's services, data and priorities. But some opportunities turn up almost everywhere. Here are six common use cases from one recent council assessment:
Inbox management tops the list for a reason. Shared mailboxes are one of the most common hidden systems in any council: a team inbox, a set of rules, and a lot of people reading, forwarding and copying information by hand. Triage, routing and a drafted first reply take a big chunk of that work away, in almost every service.
Safeguarding triage shows how AI supports professional judgement rather than replacing it. AI summarises and prioritises referrals. The front door team still makes every decision, with more time for the cases that need them most.
What "up to 200" looks like
These six are a small sample. A typical assessment surfaces up to 200 opportunities. Some come from our catalogue of 250+ proven use cases. Many come from the council's own teams in the workshops.
Derby City Council was our first assessment, and our broadest. It identified 261 opportunities across adult social care, children's services, customer service, and internal and staff services. They ranged from library enquiries and blue badge requests to care review planning, FOI requests and fleet optimisation. We then narrowed that down to 54 to take forward.
The bit that matters is the narrowing. A long list of ideas is easy. A prioritised, costed set that a council can actually deliver is what makes the difference.
Seeing it work
At the end of every assessment, we show a working demo of the kind of use case it has surfaced. In the webinar, I used a housing compliance example.
A housing officer uploads the property register and a batch of gas safety certificates.
AI reads each certificate, checks it against the register.
It flags anything missing, expired or inconsistent.
It drafts the follow-up letters for the officer to review and send.
Today, that work is usually done by hand: opening each certificate, checking dates and addresses, updating a spreadsheet and writing to tenants or contractors one at a time. It's exactly the kind of hidden system I described above, and exactly where AI gives hours back.
One business case, not one tool at a time
The last change is the business case. Instead of costing each use case separately, we now bring every opportunity, across every service, into one consistent calculation. That means options can be compared like for like, and each benefit comes with a clear plan for how it’ll be realised.
It’s built on the council's own data and checked by its own teams. For leaders, that's what turns a list of interesting ideas into a decision they can take to committee or board.
Where to start looking
If you want a quick test of your own, pick one piece of work and ask the team to walk you through how it actually gets done, end to end. Count the spreadsheets, shared mailboxes and rules it passes through. That’s actually where the hidden system is, and where the time goes.
If you suspect your council has a few systems that aren't systems, or a few sofas nobody can claim, we'd be glad to come and talk to your SLT about what an assessment could find.
Blog series:
Read the previous blog in the series: Inside an assessment - five days on site, every service, every level.
Next in the series: Every number checked - from assessment report to a business case finance will back, by Aneeq Star.
Comments